<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Mahdi Nasseri on Medium]]></title>
        <description><![CDATA[Stories by Mahdi Nasseri on Medium]]></description>
        <link>https://medium.com/@mahdinasseri?source=rss-ee8cd6238135------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*33321Mfsi5UMOVwKcX1-pA.jpeg</url>
            <title>Stories by Mahdi Nasseri on Medium</title>
            <link>https://medium.com/@mahdinasseri?source=rss-ee8cd6238135------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 14:13:00 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mahdinasseri/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The End of the Data Analyst, The Beginning of the Age of Data Analytics]]></title>
            <link>https://medium.com/@mahdinasseri/the-end-of-the-data-analyst-the-beginning-of-the-age-of-data-analytics-3f5f0c6a61e4?source=rss-ee8cd6238135------2</link>
            <guid isPermaLink="false">https://medium.com/p/3f5f0c6a61e4</guid>
            <category><![CDATA[generative-ai-market]]></category>
            <category><![CDATA[career-paths]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[future]]></category>
            <dc:creator><![CDATA[Mahdi Nasseri]]></dc:creator>
            <pubDate>Wed, 18 Nov 2026 10:11:53 GMT</pubDate>
            <atom:updated>2026-11-18T10:11:53.528Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5h5OICIEFHV0AIXTuww0_Q.jpeg" /><figcaption>Photo by Ghazal Gh on Unsplash</figcaption></figure><h3>Data Analysts: Standing at the Threshold</h3><p>Back in 2023, when the first sparks of ChatGPT were beginning to light up the world, I was working as a Data Analyst at Divar. At the time, there were nearly 35 other Data Analysts working at the company. We held a meeting to discuss the <strong>future of data analytics in the age of GenAI</strong>.</p><p>Even back then, I was convinced that within a year or two, none of us would be Data Analysts anymore. My prediction was a little pessimistic. We are still Data Analysts, but as far as I know, fewer than five Data Analysts remain across the entire company today. Yet these days, I feel even more confident repeating that prediction.</p><p><strong>The day of reckoning for Data Analysts has arrived.</strong></p><p>Reality is finally catching up, although a little later than I expected. It feels like the first sparks and smoke rising before a young volcano erupts. We are now in 2026, and over the past six months, we have seen even stronger signs of this transformation.</p><p>The Data Analyst role, as we have known it for years, is undoubtedly approaching a major turning point. We still don’t know exactly what lies beyond this threshold. But we can see that our work and professional identity may no longer be defined by the title of Data Analyst.</p><p>Today, I lead a Data Analytics team at SnappMarket. Over the past six months, my team and I have rolled up our sleeves to bring about a major transformation. A transformation that begins with the gradual decline of the Data Analyst as a standalone organizational role and leads somewhere we cannot yet fully picture.</p><p>On one hand, this change may gradually integrate Data Analyst responsibilities into other organizational roles. On the other, it may allow data analytics to become far more deeply embedded across every organizational process.</p><p><strong>We decided not to sit around waiting for someone else to build the future. We decided to move forward and build it ourselves.</strong></p><h3>The Data Analyst Is Becoming History, but Data Analytics Is Here to Stay</h3><p>Data analytics is becoming an increasingly accessible and widely available capability. Many activities that required specialized expertise, expensive tools, and skilled professionals just a few years ago can now be performed with AI at a fraction of the time and cost.</p><p>Skills that once gave individuals or organizations a competitive advantage are gradually becoming baseline capabilities. This is what we mean by <em>commoditization</em>. When an activity becomes widely accessible at a relatively low cost, simply being able to perform it is no longer enough to create a sustainable competitive advantage.</p><p>Working with data sources, writing SQL queries, understanding basic data analytics concepts, programming in Python, knowing your way around Business Intelligence tools, and even implementing some advanced machine learning models such as forecasting, clustering, and recommendation systems are gradually becoming less of a competitive advantage. Companies will no longer pay Data Analysts high salaries simply because they possess these skills.</p><p><strong>The economic value of these skills as standalone areas of expertise and competitive advantages is declining.</strong></p><p>That said, we will still need specialists and processes to design data infrastructure, ensure data quality, manage complexity, and solve specialized problems. But simply being able to perform activities that were once considered specialized will no longer guarantee our professional value.</p><p>If you find this hard to believe, try connecting Claude, Codex, or Cursor to your data and analytics sources, regardless of their format or level of structure. You will get a glimpse of this transformation firsthand.</p><p><strong>Many of the tasks that defined our expertise yesterday can now be performed with surprising ease.</strong></p><p>Hearing this from someone who has spent more than ten years in this field may sound unsettling, strange, or even incomprehensible to someone just starting their career or somewhere along the way. But we need to prepare ourselves for this reality.</p><p>This does not mean data analytics is disappearing. Quite the opposite. The process is becoming so much cheaper and more accessible that it will gradually find its way into the heart of every organizational process: decision-making, development, improvement, maintenance, and innovation.</p><p>It will become so deeply intertwined with other activities that separating it as an independent function, with its own standalone value proposition, will become increasingly difficult.</p><p><strong>Data analytics is not going away. It is becoming embedded in every organizational process, and in that sense, it is here to stay.</strong></p><p>As Data Analysts, we can take pride in having played an important role in the evolution of data-driven decision-making. We helped build the technological foundations, tools, and methodologies that technology can now make available to organizations faster, cheaper, and at a much greater scale.</p><h3>So, What Should We Do Now?</h3><p>The trends shaping the future of data analytics have important implications that can be divided into two broad categories.</p><p>The first concerns Data Analysts: people who have built their careers around data analytics.</p><p>The second concerns employers: organizations that hire Data Analysts.</p><p>The first group needs to answer a fundamental question: What should we do to continue creating value and earning a living in the future?</p><p>The second needs to figure out how to embed data analytics capabilities into organizational decision-making processes and benefit from them.</p><h3>For Data Analysts</h3><p>The first and easiest option (and, in my opinion, the worst one) is to change your career path.</p><p>Why do I think this is the worst decision? Because moving into another technology career does not necessarily put you in a better position.</p><p>Every career path in technology is facing automation and the replacement of human activities to some degree. Product management, software development, maintenance, infrastructure, and various engineering disciplines are not immune to these changes.</p><p>So moving into another field simply to escape this wave is probably not the right choice. Especially when you are starting from scratch and have to compete with experienced professionals who are also equipped with AI tools.</p><p>The last option, and perhaps the most complicated and unusual one, is to leave the technology industry altogether and get as close as possible to the physical realities of customers’ lives.</p><p>If plumbing, carpentry, or cabinetmaking isn’t your thing, at least consider moving closer to the operational layers that directly serve customers’ needs. This could mean working in value delivery and direct customer interaction, or building small and medium-sized businesses that address specific needs for specific customer groups in particular geographic areas, such as sports services or logistics solutions.</p><p>Of course, this path is not necessarily right for everyone either.</p><p>But the important point is that the closer we get to where value is actually created and delivered to customers, the more we encounter problems and complexities that cannot be solved simply by automating knowledge work.</p><p>Neither of these options is what I am choosing right now.</p><p>I am looking for a third path, one that helps me, as a Data Analyst, maintain and expand my ability to create value for organizations here and now:</p><p><strong>Moving up the value chain of decision-making.</strong></p><p>What we did as Data Analysts was only a relatively small part of the broader value-creation process within organizational decision-making.</p><p>An organization is, fundamentally, an institution for making decisions. Organizations need to make sound and rational decisions to reduce risk, increase efficiency, and ultimately create valuable and affordable solutions for their customers.</p><p>To do this, data must be collected from the surrounding environment, organizational operations, and other sources. It must go through processes of cleaning, standardization, and storage, and become available at the right time. Then, using statistics, mathematics, critical reasoning, and analytical methods, it can be examined to generate insights that support decision-making.</p><p>But that’s not the whole story.</p><p>There are several important steps between having an insight and turning it into actual value. A decision must be made, translated into action, and ultimately produce the expected value in the real world.</p><p>So <strong>what we, as Data Analysts, were really doing was contributing to the decision-making process, not simply working with data.</strong></p><h3>The Value Creation Chain in Decision-Making</h3><p>We can break down the value creation chain in decision-making into four broad stages:</p><p><strong>1. Data Work</strong></p><p>All the technical skills and activities related to data, including collection, storage, cleaning, processing, retrieval, modeling, and visualization.</p><p><strong>2. Decision Making</strong></p><p>All the analytical skills and knowledge required to understand problems, interpret data, reduce uncertainty, evaluate scenarios, and recommend decisions.</p><p>Descriptive, exploratory, predictive, and prescriptive analytics all belong to this stage.</p><p><strong>3. Decision Execution</strong></p><p>Turning decisions into real actions within the organization.</p><p>This includes designing execution processes, allocating resources, coordinating across teams, managing change, and ensuring that decisions are properly implemented.</p><p><strong>4. Value Realization</strong></p><p>Ensuring that the actions taken actually deliver the expected value.</p><p>This includes measuring economic and operational outcomes, evaluating the impact of decisions on customers, assessing results, and learning from them.</p><p>This distinction matters because executing a decision does not necessarily mean realizing its intended value.</p><p>An organization might make the right decision and execute it correctly, yet fail to achieve the expected outcome because of changing market conditions, competitors’ reactions, or flawed initial assumptions.</p><p>Similarly, value realization is not simply about completing an action successfully in the short term. We need to make sure that the value created is real, measurable, and sustainable to the greatest extent possible.</p><h3>Organizational Maturity and Where Value Is Created</h3><p>For each of these stages, we can identify three broad conditions that describe the strategic importance and competitive value of a capability:</p><p><strong>1. Commodity</strong></p><p>A capability that has gradually become standardized and widely available at a relatively low cost. In this situation, simply possessing that capability does not create a meaningful competitive advantage.</p><p><strong>2. Differentiator</strong></p><p>A capability that allows an organization to outperform its competitors and create a competitive advantage. Organizations are willing to invest in developing and maintaining such capabilities.</p><p><strong>3. Bottleneck</strong></p><p>A capability whose capacity or limitations constrain the growth and performance of the entire value chain. Whoever manages to remove that constraint can create significant value for the organization.</p><p>These three conditions are not necessarily mutually exclusive. A capability can become a commodity in the broader market while remaining the primary bottleneck to growth within a particular organization.</p><p>For example, data analytics tools may have become commodities at the market level. But in an organization with unreliable and fragmented data, data quality and integration may still be the biggest bottlenecks to effective decision-making.</p><p>Therefore, the model I use to describe the future career of Data Analysts should not be seen as a universal rule for how all organizations will evolve.</p><p>It is better understood as an <strong>organizational maturity model</strong>. It illustrates how, as organizations become more mature in their use of data and technology, the main sources of competitive advantage and the bottlenecks in their value chains may shift.</p><p>And these are precisely the areas where I believe Data Analysts should increasingly focus their work.</p><p>The following table represents my perspective on this transformation across three levels of organizational maturity:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tMo50pavp2GVCu3n7QYj5g.png" /><figcaption>The Data Analyst’s Strategy for Moving Up the Value Creation Chain</figcaption></figure><p>In the past, Data Work itself was either a competitive differentiator or a bottleneck to organizational growth.</p><p>People who knew how to install databases, clean data, load it into storage, and eventually deliver it in Excel tables and spreadsheets, accompanied by a few charts and a couple of paragraphs of interpretation, created significant value for organizations. As a result, they were well paid.</p><p>Even today, organizations that have fallen far behind in technology are still in this situation.</p><p>But with the development of data infrastructure, a wide range of databases, and increasingly affordable Business Intelligence tools, Data Work alone no longer creates a meaningful competitive advantage.</p><p>In recent years, organizations have been looking for people who could <strong>understand</strong> the data and answer the question: “So what?”</p><p>That was Decision Making.</p><p>Organizations that were better than their competitors at this stage gained an advantage. And those who could do this work, Data Analysts, enjoyed a distinctive professional position.</p><p>Yet even during that period, the main bottleneck for many organizations was turning analysis into action and realizing value.</p><p>Businesses struggled to translate what they had learned from their data into actual operational changes and, ultimately, economic value.</p><p>But the future may look different.</p><p>With the help of AI, a significant portion of the cognitive and analytical work required for decision-making can now be performed at a speed and scale that organizations could not previously achieve.</p><p>For many business questions, we no longer need to wait days for data extraction, dashboard development, and an analyst’s interpretation.</p><p>These days, when our AI-powered data assistant can build a comprehensive dashboard in just a few minutes based on a question raised during a meeting, analyze it, and even suggest possible solutions, we begin to understand the enormous potential now available to us.</p><p>We should not forget that producing a good analysis does not necessarily lead to a good decision. Recommending a solution does not mean fully understanding all the constraints and complexities of a business.</p><p>Identifying the right problem, understanding the business context, evaluating the consequences of a decision, and taking responsibility for it still matter.</p><p>Nevertheless, a significant portion of what once gave Data Analysts a competitive advantage is becoming a widely accessible capability.</p><p>Under these conditions, a business’s competitive advantage will increasingly depend on its ability to execute decisions and realize value through its operations.</p><p>And that is exactly where you and I, as Data Analysts, need to move.</p><h3>Who Is the Data Analyst of the Future?</h3><p>If we are going to move beyond our previous professional identity, we need to understand what should replace it.</p><p>In my view, the Data Analyst of the future is not simply someone who knows more tools or can implement increasingly sophisticated analytical models.</p><p>These skills will remain useful, but they will no longer be enough to define our professional identity.</p><p>We need to gradually move from being specialists in working with data to becoming specialists in problem-solving and value creation.</p><p>I believe several key capabilities will become increasingly important in this transition:</p><p><strong>1. Business Acumen</strong></p><p>We need to understand the business, not just its data.</p><p>Understanding the business model, unit economics, customer behavior, cost structure, supply chain, competitors, and the main drivers of value creation will no longer be secondary skills for Data Analysts. They will become part of our core expertise.</p><p><strong>2. Decision Intelligence</strong></p><p>We need to be able to identify the right problems, reduce uncertainty, challenge assumptions, evaluate different scenarios, and help decision-makers make better decisions.</p><p>Our value should not be limited to answering questions that others ask. We should be able to identify more important questions ourselves.</p><p><strong>3. AI Orchestration</strong></p><p>We need to learn how to use AI as an integral part of our work.</p><p>This means understanding how to connect intelligent agents and tools to data sources, evaluate their outputs, assess the quality and reliability of their results, and use them to expand our problem-solving capabilities.</p><p>The ability to work with AI will gradually move from being an advantage to becoming a necessity.</p><p><strong>4. Operational Excellence</strong></p><p>We need to get closer to what actually happens in the real world of business.</p><p>Understanding operational processes, resource constraints, execution mechanisms, change management, and how to collaborate with teams responsible for implementing decisions will become increasingly important.</p><p><strong>5. Strategic Thinking</strong></p><p>We need to distinguish important problems from less important ones, prioritize value-creation opportunities, and understand where an organization should invest its limited resources.</p><p>This means moving beyond waiting for analytical requests and actively participating in identifying the organization’s most important opportunities and challenges.</p><p>Not every Data Analyst needs to become an expert in all these areas.</p><p>But I believe that the more we expand our knowledge and skills beyond Data Work and toward these capabilities, the better our chances of maintaining a valuable professional position in the future.</p><p>This transformation is not simply about learning a few new skills.</p><p><strong>It is about redefining the value we create for organizations.</strong></p><p>We need to move from people who answer business questions to people who participate in identifying problems, shaping decisions, and realizing value.</p><p>At SnappMarket, we have designed a specific path to make this transition. We know, of course, that no one can predict the future with certainty.</p><p>But one thing I can say with confidence right now is this:</p><p><strong>You need to get up and get moving.</strong></p><h3>For Employers</h3><p>This is absolutely not the time to eliminate Data Analyst positions in your organization, say goodbye to your analysts, and then open ChatGPT thinking that people like me have declared the Data Analyst dead.</p><p>That would be the worst possible move.</p><p>We need to distinguish between two different concepts: <strong>the Data Analyst as a standalone job role, and data analytics as an organizational capability.</strong></p><p>The number of standalone Data Analyst positions may decline in the future. Some of their responsibilities may become integrated into other organizational roles, and the nature of the job itself may change significantly.</p><p>But this does not mean that the importance of data analytics as a capability will decline.</p><p>Quite the opposite. Data analytics will become more deeply embedded than ever in the everyday activities of organizations.</p><p>The truth is that we still do not know exactly what the best way to use AI in decision-making and value creation looks like.</p><p>You need to give your teams, especially your Data Analysts, the freedom and confidence to build this path themselves.</p><p>Because of their understanding of data, analytical processes, and business problems, they may be among the best people to lead this transition.</p><p>The real magic lies in <strong>Integration and the Context Layer</strong>.</p><p>If you don’t have the necessary infrastructure, get started quickly.</p><p>First, make sure your data accurately reflects what is happening in your business. You need <strong>Trusted Data</strong>.</p><p>Then, build the infrastructure to integrate your data with AI tools. At the same time, you need to make sure your business context is properly communicated to AI.</p><p>Hopefully, you have already documented at least some of your organizational knowledge.</p><p>That’s where you should start. And if you haven’t done so yet, that’s still where you should start.</p><p>Now is the time to systematize decision-making.</p><p>Look closely at your organizational processes. Understand how people and systems make decisions at different points, and identify which parts of these processes can be enhanced using new AI capabilities.</p><p>The real magic happens when you can transform data analytics from a separate, specialized activity into a capability that runs throughout the entire organization.</p><p>I hope to share a more structured and tested account of the journey we have started at SnappMarket here in the future.</p><p><strong>We are still at the beginning of the journey ourselves.</strong></p><h3>The World Is Changing at an Incredible Pace</h3><p>This may be the most repeated statement you hear these days. Yet it is also one of the biggest truths you can hear: the world is changing at an incredible pace, and it waits for no one.</p><p>The time for transformation has come. We need to get up and move.</p><p>As someone who has spent much of his professional life in data analytics, I now believe we need to burn in the ashes of our own fire, like the phoenix, and be reborn.</p><p><strong>Perhaps it is time to accept the end of the Data Analyst as we know it, so we can help bring about the beginning of a new age of data analytics.</strong></p><p><em>Author’s note: This article was originally written in Persian and translated into English with the assistance of AI.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3f5f0c6a61e4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How I Entered the World of Data from a Different Path (and You Can Too)]]></title>
            <link>https://medium.com/@mahdinasseri/how-i-entered-the-world-of-data-from-a-different-path-and-you-can-too-3751d4ef5292?source=rss-ee8cd6238135------2</link>
            <guid isPermaLink="false">https://medium.com/p/3751d4ef5292</guid>
            <category><![CDATA[data-analyst]]></category>
            <category><![CDATA[career-change]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[career-advice]]></category>
            <category><![CDATA[data]]></category>
            <dc:creator><![CDATA[Mahdi Nasseri]]></dc:creator>
            <pubDate>Tue, 11 Nov 2025 07:01:57 GMT</pubDate>
            <atom:updated>2025-11-11T07:43:13.628Z</atom:updated>
            <content:encoded><![CDATA[<p>More than a decade ago, <em>Harvard Business Review</em> published a headline that said: <strong>“Data Scientist: The Sexiest Job of the 21st Century.”</strong> (<a href="https://hbr.org/2012/10/data-scientist-the-sexiest-job-of-the-21st-century">link</a>)</p><p>Since then, many people have tried to change their career path toward data, in any way they could. But now, almost 13 years later, can we still confidently say that a career in data has a bright future? And if someone wants to switch to it now, what should they do?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*1jW4rbSC5asSSUgc" /><figcaption>Photo by <a href="https://unsplash.com/@saulomohana?utm_source=medium&amp;utm_medium=referral">Saulo Mohana</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Back in 2014, before I seriously entered the data world, I was the marketing manager of an online photo printing company called <a href="https://axprint.com/"><em>Axprint</em></a>. That’s where my curiosity about <em>“</em><strong>why some decisions work and others don’t</strong><em>”</em> gradually pulled me toward data.</p><p>I soon realized that <strong>having analytical, data-driven insight into consumer behavior</strong> could become an unbeatable competitive advantage in my career.<br>So, after three years, I resigned and my transition to the world of data began around 2017, right when data analytics was becoming a hot career both in Iran and around the world.</p><p>But what about now? For someone considering such a move today, is it too late? Is data science still the sexiest job in tech? Isn’t the market already saturated? And with the rise of GenAI, do we still need people who can actually <em>work with data</em>? Won’t AI take over?</p><p>I don’t have a clear answer to these questions. I’m concerned about my own career path too, but I’m not planning to leave it anytime soon and that’s exactly why I can confidently recommend it to anyone who has the right mindset and abilities.</p><p><strong>I believe that in any possible future, the person who knows how to understand and use data will always have the upper hand.</strong></p><p>This post is not meant to be a <em>step-by-step roadmap</em> to learning data. You can find that in seconds through Google or ChatGPT. Instead, this article is written for people who want to change their career path from another field into data. It’s based on my own journey and on helping others do the same over the years.</p><h3>1. First, understand what “data” really means.</h3><p>The best way to understand data is by <strong>working with it</strong>. That’s my first and most important advice. Wherever you are, and whatever data you have access to, don’t hesitate roll up your sleeves and dive in.<br>You’ll soon realize what massive piles of structured (numerical) and unstructured (textual) data truly are.</p><p><strong>Data is a quantitative understanding of a qualitative world and qualitative action (decision-making) based on quantitative reasoning.</strong></p><p>Exactly what <em>Sherlock Holmes</em> used to do! If you’ve ever enjoyed his stories, the world of data will fascinate you.</p><p>But depending on where you stand in this world, it takes different forms.<br>It’s worth researching the three main paths in the data field:</p><ul><li><strong>Data Analysis: </strong>finding patterns and answering real business questions</li><li><strong>Data Science: </strong>modeling, prediction, and applying statistics &amp; machine learning</li><li><strong>Data Engineering: </strong>building the infrastructure that makes all of it possible</li></ul><p>Ten years ago, these paths weren’t so distinct but now they are. Despite their overlap, each has its own requirements. None is superior to the others; your <strong>interest</strong> and <strong>strengths</strong> should decide.</p><p>I personally started from data analysis, because my focus was never the data itself, it was decision-making based on data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*MPS2QB2x3Gcr99aX" /><figcaption>Photo by <a href="https://unsplash.com/@keiteu_ko?utm_source=medium&amp;utm_medium=referral">Keiteu Ko</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>2. Your past skills are gold, not a weakness.</h3><p>Working with data is all about <strong>solving problems with a data-driven mindset</strong>.</p><p>The data world needs people who truly understand the <em>context</em> of problems.<br>Someone who can interpret the meaning behind every number or behavior is far more valuable than someone who just runs algorithms.</p><p>So, whatever your background — marketing, operations, development, or customer service — your ability to solve problems will serve you well in data. Don’t make the mistake I did: assuming you must start from zero. <strong>Data is never separate from its context</strong>. You’re not on step zero; you’ve already climbed part of the path.</p><p>I came from marketing. At first, I thought all my marketing knowledge was useless. But when I started working on real projects, I realized how deeply it helped me understand and frame data-driven solutions.</p><p>My understanding of consumer behavior, marketing concepts, business development, team management, campaign optimization, stakeholder communication, and strategic planning, all became valuable assets in my data career.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*HuxK4mgEOni4DqUo" /><figcaption>Photo by <a href="https://unsplash.com/@neom?utm_source=medium&amp;utm_medium=referral">NEOM</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>3. Learn tools with purpose.</h3><p>Tools don’t win the race but even <em>Schumacher</em> couldn’t win without a car.</p><p>If you want to work in data, be ready to get your hands dirty.</p><p>Data comes in countless formats and from various sources; tools are equally diverse. Your main skill is not mastering every tool, but <strong>adapting quickly and using each tool effectively</strong>.</p><p>Data analysis isn’t just Power BI or Looker Studio. You need to know databases, at least one or two programming languages, and some key libraries. Get familiar with Google Analytics and other analytical services.</p><p>Here are the four essential tools I recommend for starting:</p><ul><li><strong>Excel or Google Sheets</strong></li><li><strong>SQL</strong></li><li><strong>Python (or R)</strong> — especially Pandas, Matplotlib, and core AI/ML libraries</li><li><strong>Power BI or Tableau</strong></li></ul><p>I knew some programming from university but learned Python from scratch.</p><p>I had to relearn SQL. I used to hate Excel, now I love it. My first dashboards were made in Tableau, with YouTube tutorials in thick Indian accents. Later I learned Power BI, Google Analytics, and more advanced systems as my projects grew.</p><p>Eventually, at Snappmarket, my team even set up a full in-house analytics infrastructure on our Linux servers, with automated data jobs integrated into marketing operations.</p><p><strong>And the learning never stops.</strong></p><p>Now, I know I must learn GenAI-based tools or I’ll slowly fade out of the field.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*oaPquHSVDE3TNFQV" /><figcaption>Photo by <a href="https://unsplash.com/@i_abhinand?utm_source=medium&amp;utm_medium=referral">Abhinand Venugopal</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>4. Learn the computer science fundamentals no one talks about.</h3><p>If you studied computer science, you already know this part.</p><p>But most people switching from other fields don’t, and that’s a huge gap. Understanding basic computational concepts is essential for every data professional.</p><p>The most important ones are:</p><ol><li><strong>Algorithm design</strong></li><li><strong>Database fundamentals</strong></li><li><strong>Data structures</strong></li><li><strong>Statistics and probability</strong></li><li><strong>Discrete mathematics</strong></li></ol><p>You don’t need to master them like a CS major, but you should know the basics.</p><p><strong>And please learn them from books, not videos or blogs. Online sources can confuse you. There are great beginner-friendly textbooks for each topic.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*zP9qq717BpOr2c75" /><figcaption>Photo by <a href="https://unsplash.com/@leonjaywu?utm_source=medium&amp;utm_medium=referral">Leon Wu</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>5. Projects matter more than certificates.</h3><p>No one will hire you just because of a certificate.</p><p>Data is a company’s most valuable asset and they won’t hand it to someone who just “took a course.”</p><p>But when you have visible, tangible projects — things people can see, test, and discuss — the story changes.</p><p>Before getting your first data job, work on as many volunteer or self-initiated projects as you can. Publish them on <a href="https://github.com/">GitHub</a> or write blog posts about them. Talk about them proudly. It works. I promise.</p><p>Finding data is easy. The internet is full of public datasets. Download them, analyze them, build a narrative around them, and share your insights.<br>Join <a href="https://www.kaggle.com/">Kaggle</a> challenges, and keep an eye on open data competitions from startups. Use <a href="https://www.datacamp.com/">DataCamp</a> or <a href="https://colab.research.google.com/">Colab</a>, and connect your life to Git. Keep enriching your profile day by day.</p><p>That’s exactly how I started with public datasets, challenges from <a href="https://github.com/mahdinasseri/Ads-Networks-Performance-and-Fraud-Analysis">Bazaar</a>, <a href="https://vrgl.ir/fmkjn">Poosheh</a>, and <a href="https://vrgl.ir/8SPHT">Aparat’s</a> annual reports. I wrote <a href="https://virgool.io/@mahdi/list/kfz7ptxv3ti8">blog posts</a>, shared my findings, and built my professional identity that way. Every job I’ve ever landed came from that process.</p><p><strong>Being active in professional communities truly shapes your career.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*678S68-ezdNukM00" /><figcaption>Photo by <a href="https://unsplash.com/@praveentcom?utm_source=medium&amp;utm_medium=referral">Praveen Thirumurugan</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>6. Where to find your first real-world experience</h3><p>Real projects are priceless, real problems, real data, real decisions, real impact.</p><p>Your first real experience might be imperfect, but it’s the most valuable one.</p><p><strong>If you still have a non-data job, start <em>right there</em>. Look at your company’s data.</strong></p><p>Study the reports, ask questions, and use data in your current tasks even in small ways. That’s where your first real project begins.</p><p>Gradually ask your manager for more data-related responsibilities. It doesn’t matter if you’re in product, sales, marketing, or support, start solving your daily problems with data.</p><p>Show your results, and once you prove your ability, ask for an official data role.</p><p>Then, activate your professional network. Tell friends and colleagues you’re looking for data projects.</p><p>Announce it on LinkedIn. When a startup offers a collaboration, take it, even if it’s unpaid at first.</p><p>Startups are hungry for insights but can’t afford full-time analysts.</p><p>My first data projects began right inside Axprint’s marketing team.<br>Later, through my wife who worked at <a href="https://www.filimo.com/"><em>Filimo</em></a>, I met its founder, <a href="https://www.linkedin.com/in/mjshakouri/">Mohammad-Javad Shokouhi</a>, who invited me to analyze their data. I dropped everything and did it.</p><p>Then, <strong>the late Mohammad Zahedi</strong> from <a href="https://eventcenter.ir/"><em>EventCenter</em></a> gave me their data to analyze, and things snowballed from there. These were my first real-world experiences and I gave them everything I had.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*vdeDa4vHelyBOVXV" /><figcaption>Photo by <a href="https://unsplash.com/@jakubzerdzicki?utm_source=medium&amp;utm_medium=referral">Jakub Żerdzicki</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>7. Don’t fear being a “junior.”</h3><p>Honestly, we’re <em>all</em> juniors now. With the speed of change in today’s tech world, no one can escape that.</p><p>The key is continuous learning and the courage to embrace uncertainty.</p><p>So if you’re switching to data, don’t fear being new. Instead of chasing vertical promotion, focus on <strong>decision-making and impact</strong>.</p><p>Each month, ask yourself: <em>If I weren’t here, what decision wouldn’t have been made or would have been made differently?</em></p><p><strong>A data analyst grows not by title, but by influence.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*5qxUr9rGGtA0wNYk" /><figcaption>Photo by <a href="https://unsplash.com/@mufidpwt?utm_source=medium&amp;utm_medium=referral">Mufid Majnun</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>8. If you read only one book, make it this one.</h3><p>I’ve recommended this book to everyone even slightly interested in data.</p><p><strong>You can call it the <em>Bible</em> of data analysts or simply, <em>the mathematics of business</em>.</strong></p><p>It teaches you how to view business problems through the lens of data.</p><p>No matter your role, analyst, marketer, developer, or manager, this book will push your thinking several steps forward:</p><p>📘 <strong>Lean Analytics: Use Data to Build a Better Startup Faster (</strong><a href="https://leananalyticsbook.com/"><strong>link</strong></a><strong>)<br></strong>by <a href="https://www.linkedin.com/in/alistaircroll/">Alistair Croll</a> and <a href="https://www.linkedin.com/in/byosko/">Benjamin Yoskovitz</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/0*z4HE75-NzEa7hqA0" /></figure><h3>Career switching takes time but it’s possible.</h3><p>No one becomes a Data Analyst or Data Scientist overnight. If it were easy, it wouldn’t be valuable.</p><p>So give it time, anywhere between 3 to 12 months, depending on your background. Stay focused, learn consistently, and most importantly, <em>learn by doing.</em></p><p>In my case, despite a tech background, it took me about three months to land my first professional project, and a year later my data career was fully running.</p><p>I was lucky to dedicate those six months full-time. You might not have that chance. So be smarter with your time. Use your “dead hours,” and remember: the only real way to learn data is <strong>by working with it</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ChP7BCpgxe6lZwvk" /><figcaption>Photo by <a href="https://unsplash.com/@sfkopstein?utm_source=medium&amp;utm_medium=referral">shraga kopstein</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><blockquote>The world of data is made for curious, analytical minds who love to learn. For those who see data not just as a job but as a clearer way to see the world.</blockquote><p><strong>Postscript:<br></strong>This article was originally written in Persian and translated into English with the assistance of ChatGPT, preserving the author’s original tone and narrative style.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3751d4ef5292" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Our Coexistence with AI in Data Analysis (LLM Workplace Guideline)]]></title>
            <link>https://medium.com/@mahdinasseri/our-coexistence-with-ai-in-data-analysis-llm-workplace-guideline-f654f9eb190a?source=rss-ee8cd6238135------2</link>
            <guid isPermaLink="false">https://medium.com/p/f654f9eb190a</guid>
            <category><![CDATA[llm]]></category>
            <category><![CDATA[guidelines]]></category>
            <category><![CDATA[genarative-ai]]></category>
            <category><![CDATA[ai-governance]]></category>
            <category><![CDATA[data-governance]]></category>
            <dc:creator><![CDATA[Mahdi Nasseri]]></dc:creator>
            <pubDate>Thu, 02 Oct 2025 06:33:43 GMT</pubDate>
            <atom:updated>2025-10-02T06:33:43.659Z</atom:updated>
            <content:encoded><![CDATA[<p>These days in our marketing team, Artificial Intelligence and Large Language Models (GenAI/LLM) are almost present in all steps of the data analysis cycle: from problem finding and problem definition to framing, implementation, final analysis, and even data storytelling. These tools increase both speed and quality of work, and sometimes they act like an “all-knowing assistant.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*9ys1mBCdY_Obd6w2" /><figcaption>Photo by <a href="https://unsplash.com/@all_about_teng?utm_source=medium&amp;utm_medium=referral">Hồ Xuân Hoàng Dương</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>But together with these opportunities, important questions also appear:</p><ul><li>Can AI make mistakes?</li><li>Can our data become unsafe?</li><li>Can it bias us toward existing methods?</li><li>Can it slowly make us lazy and move us away from our own skills?</li><li>What is really our role, and which skills should we develop?</li></ul><p>These concerns are not only ours. Almost everywhere in the world people are facing the same questions. At the level of law and data governance, regulators have not yet reached a final “best practice.” Some large companies, especially those closer to governments or with more traditional structures, have banned the use of ChatGPT inside their organizations and even blocked it on internal networks. Still, many employees use these tools with their personal devices; after all, who would not like to have a fast and easy assistant that helps to do things better?</p><p>On the other hand, many leading organizations have adapted quickly: they built their own internal LLMs, encouraged employees to use them, and created new processes and solutions with their help. Many others still have no clear position and prefer to wait and see which successful models will appear in the world.</p><p>The reality is that challenges of data governance, productivity, and employee career paths are still not fully clear. However, we decided not to wait passively but to create a temporary framework for using this technology. This framework has turned into a code of conduct for using such tools: a set of do’s and don’ts, principles, red lines, and useful recommendations.</p><p>Now, I decided to share this guideline of using ChatGPT in data analysis processes of our marketing team with you. You can use it as it is, or give suggestions for improving and upgrading it.</p><p><a href="https://mahdinasseri.github.io/llm-workspace-guidelines/guidelines/llm-workspace-guidelines-en.html">Guideline for Using GenAI in Data Analysis of the Marketing Team</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f654f9eb190a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Golden Question in Data Analyst Interviews (based on +100 interviews over the past 5 years)]]></title>
            <link>https://medium.com/@mahdinasseri/my-golden-question-in-data-analyst-interviews-based-on-more-than-100-interviews-over-the-past-5-7711b8843d56?source=rss-ee8cd6238135------2</link>
            <guid isPermaLink="false">https://medium.com/p/7711b8843d56</guid>
            <category><![CDATA[interview]]></category>
            <category><![CDATA[interview-questions]]></category>
            <category><![CDATA[hiring-strategy]]></category>
            <category><![CDATA[data-analyst]]></category>
            <category><![CDATA[team-building]]></category>
            <dc:creator><![CDATA[Mahdi Nasseri]]></dc:creator>
            <pubDate>Mon, 22 Sep 2025 17:41:38 GMT</pubDate>
            <atom:updated>2025-09-22T18:33:53.003Z</atom:updated>
            <content:encoded><![CDATA[<p>After running over a hundred interviews with data analysts in the past five years, I realized there is one question that shows a person’s real ability much better than any technical test or CV:</p><blockquote><strong><em>“Imagine you’ve just joined our company and have three months to work on any issue or project you want. What would you choose, and why?”</em></strong></blockquote><p>At first glance, this question may look “soft.” But actually, it’s a very focused and precise question. It shows how the candidate thinks, sets priorities, and creates impact on the business.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Z5BhYb-PkDOzAnX8" /><figcaption>Photo by <a href="https://unsplash.com/@charlesdeluvio?utm_source=medium&amp;utm_medium=referral">charlesdeluvio</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>What I’m really looking for</h3><p>If you can answer this question well in my interview, I’ll probably hire you.</p><p>This question isn’t about guessing what I want. It’s not to check if you know our current projects.</p><p>I ask it to see how a candidate aligns their skills, curiosity, and initiative with our business context.</p><p>This simple question helps me measure a few powerful “meta-skills”:</p><p><strong>Capabilities:</strong> I expect the candidate to focus on their own strengths. Use past experience and base their problem-solving approach on what they know best.</p><p><strong>Learning mindset:</strong> From the answer, I see how motivated they are to learn new methods, face new challenges, and push themselves.</p><p><strong>Business impact and domain awareness:</strong> Can they show they understand our industry and challenges well enough to pick a project that really matters to us, not just in theory?</p><p><strong>Initiative and proactive approach:</strong> Does the answer show they’re a self-starter who actively looks for problems to solve and opportunities to use? How ready is their mind to spot business threats and opportunities and approach them with data-driven methods?</p><h3>How I go deeper</h3><p>I don’t stop at the first answer. I discuss it with them.</p><p>I play the role of a manager or stakeholder and try to challenge them.</p><p>I take their proposed project and push them with “T-shaped” questions:</p><p>I start wide with general concepts and then go deeper into part of their plan for solving the problem (the T-shape approach).</p><p>I remind them this is a simulated space — my goal is to see their thinking, not perfection.</p><p>If they drift too far, I guide them a little and watch how they adapt.</p><p>At the end, I ask them to take their chosen project to code level. This isn’t a “trap” — I just want to see their real technical ability and how they handle the messy, real parts of analysis.</p><h3>Step-by-step flow</h3><p>This sequence doesn’t just show skill. It also reveals the person’s mindset, initiative, and realism.</p><p><strong>Planning:</strong> First, I ask them to plan how they’d solve the problem. What phases would they take? Which stakeholders would they engage with? How would they schedule and measure progress?</p><p><strong>Business questions:</strong> Next, I focus on business questions. Why do you think this issue matters to our company? Why is a data-driven approach the best way to solve it? Which teams would your project involve?</p><p><strong>Analytical approach:</strong> Then I discuss — even argue — about their analytical plan. What data will you look at? How will you clean it? What metrics will you define? What model will you use and why this one, not that one? If I ask you to specify the accuracy of your answer, how would you do it?</p><p><strong>Technical execution:</strong> Finally, I go into the technical execution. I pick one of their suggestions. For example, if they said “I’d use a predictive model,” I ask them to talk more about this model. What technical challenges would it have? I even ask questions at the coding level.</p><h3>Red flags</h3><p>I watch out for candidates who only bring generic or “textbook” ideas — like doing segmentation just for the sake of segmentation — without linking it to real business outcomes. This shows they may not yet be ready to translate academic knowledge into practical impact.</p><p>People who only focus on routine issues of a business also score low. I expect the candidate to use this chance to show how they apply data-driven methods to identify and solve unknown problems in the business.</p><p>Candidates who don’t consider our business model in their problem choice also score low. It’s obvious that if you’re interviewing with a company, you should have studied it beforehand and know its business model, market, and competitors.</p><h3>Why this works</h3><p>By framing the interview around one open but structured question, I can see how someone thinks, designs, and executes — not just how they answer test questions.</p><p>It shows me if they’re practical, thoughtful, or a mix of both.</p><p>This method is also fair: <strong>candidates get a real, supportive simulation of how we work, and I get a true window into their skills and abilities</strong>.</p><h3>In short</h3><p>The golden question isn’t about solving a puzzle on the spot. It’s about understanding how a data analyst, when given freedom, chooses a problem, designs a solution, and makes an impact.</p><blockquote><strong>Over time, this question has become my most reliable tool for predicting success on my team.</strong></blockquote><h4>What about you?</h4><p>Have you ever been asked a “golden question” like this in an interview, or do you have your own go-to question when you’re hiring?</p><p>I’d love to hear your thoughts or experiences in the comments.</p><p><em>P.S. I used ChatGPT to help me translate this text into English.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7711b8843d56" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Bringing MACD from Wall Street to Your Dashboard: A New Way to Read E-Commerce Trends]]></title>
            <link>https://medium.com/@mahdinasseri/bringing-macd-from-wall-street-to-your-dashboard-a-new-way-to-read-e-commerce-trends-b549e324958a?source=rss-ee8cd6238135------2</link>
            <guid isPermaLink="false">https://medium.com/p/b549e324958a</guid>
            <category><![CDATA[trend-analysis]]></category>
            <category><![CDATA[kpi-dashboard]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[macd]]></category>
            <category><![CDATA[metrics]]></category>
            <dc:creator><![CDATA[Mahdi Nasseri]]></dc:creator>
            <pubDate>Fri, 18 Jul 2025 12:23:24 GMT</pubDate>
            <atom:updated>2025-07-24T04:10:22.566Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ka1Py1PGkZuNusr4" /><figcaption>Photo by <a href="https://unsplash.com/@ymoran?utm_source=medium&amp;utm_medium=referral">Yucel Moran</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>One of the most important questions that always arises when analyzing the trend of a metric over time is: <em>Is this metric growing or declining?</em></p><p>Take your daily sales as an example. Imagine you’re looking at the day-by-day changes over the past three months. Now try to answer this: <strong>What’s the overall trend?</strong> When was it growing, and when was it falling? When did the growth phase begin and when did it end? And what about the decline, when did it start, and when did it stop?</p><p>You can apply this question to any metric you like, not just sales. The core question remains the same:</p><blockquote><strong>When were we growing, and when were we declining?</strong></blockquote><p>If you’ve never analyzed the trend of such a metric in a real business before, you might suggest a simple solution:</p><blockquote><strong>Just plot the chart and look at it! If it’s sloping upward, it means we’re growing. If the curve turns downward, then we’re in decline.</strong></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*zYw31HvYcwj-vMDB" /><figcaption>Photo by <a href="https://unsplash.com/@behy_studio?utm_source=medium&amp;utm_medium=referral">Behnam Norouzi</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>While this solution is straightforward, and sometimes even useful, it has two major flaws, in order of importance:</p><ol><li>It depends heavily on the <strong><em>observer’s interpretation</em></strong></li><li>In most real-world scenarios, <em>the </em><strong><em>fluctuations are rarely obvious enough</em></strong> to judge by eye.</li><li>And often, due to the <strong>high granularity of the metric across different conditions</strong>, human observation simply isn’t feasible for meaningful trend analysis</li></ol><p>For example, take a look at the following trend:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pK_VN8wVUj0WdqHdw1Ht9Q.png" /><figcaption>Sales Trend Analysis</figcaption></figure><p>Each of us might see different growth or decline trends in the chart above, especially since there seems to be some kind of seasonal pattern in play. In such cases, sudden ups and downs can’t easily be treated as real trends.</p><p>What we need is an <strong>objective</strong>, <strong>systematic</strong>, and <strong>meaningful</strong> method for identifying the start and end points of trends, whether positive or negative. This method should help both humans and machines understand growth. In other words, it should not only be interpretable by people, but also usable in <strong>analytical computation and automation</strong>.</p><p>To tackle this problem, I turn to a method borrowed from technical analysis in the stock market. In that field, this technique is known as <strong>Moving Average Convergence Divergence</strong>, or <strong>MACD</strong>.</p><p>The core idea behind MACD is to use the <strong>difference between a short-term and a long-term moving average</strong> as an indicator for identifying growth trends.</p><p>In the following sections, I’ll explain what all of this means, and most importantly, how we can use it to solve our original question.</p><h3>From Wall Street…</h3><h4>1. Simple/ Exponential Moving Average</h4><p>A moving average smooths out the fluctuations in a metric by averaging its values over a defined time window, making it a foundational tool in trend analysis. The longer the time window, the smoother and less sensitive the curve becomes to daily changes.</p><p>There are two common types of moving averages: <strong>Simple Moving Average (SMA) and Exponential Moving Average (EMA)</strong>. SMA gives equal weight to all data points in the window, while EMA assigns more weight to recent values, making it more responsive to short-term changes. Because of this sensitivity, EMA is often preferred in technical trading, especially in fast-moving markets, as it can react more quickly to price shifts and trend reversals compared to SMA.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uCa5pb2mZMjnLPByQRv49Q.png" /><figcaption>Sales Trend and Moving Averages</figcaption></figure><p>I asked chatGPT to explain it more. If you need read it’s response in <a href="https://chatgpt.com/c/687218ba-063c-800b-b945-b9c305c30192">this link</a>.</p><h4>2. Moving Averages Crosses</h4><p>In trading, comparing short-term (e.g., 7-day) and long-term (e.g., 30-day) moving averages helps indicate momentum: if the short-term average is above the long-term one, it suggests recent growth; if it’s below, it signals decline. Key signals in technical trading arise when these averages cross:</p><ul><li>A <strong>Golden Cross</strong> occurs when the short-term average moves above the long-term average, often seen as a <strong>bullish or buy signal</strong>.</li><li>A <strong>Death Cross</strong> happens when the short-term average drops below the long-term one, typically interpreted as a <strong>bearish or sell signal</strong>.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9qfPUNU3sWXXCYYUbcG-aw.png" /><figcaption>Adding Trend Crosses to Sales</figcaption></figure><p>These crossover points are widely used by traders to identify potential shifts in market trends.</p><h4><strong>3. The Moving Average Convergence Divergence (MACD)</strong></h4><p>MACD is a momentum indicator that tracks the relationship between two moving averages of a dataset, commonly the 12-day and 26-day EMAs (Exponential Moving Averages).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rdiXoLfdi0wASUMx8aY4QQ.png" /><figcaption>Sales and MACD for Trend Analysis</figcaption></figure><p>This core MACD line, representing the difference between these two EMAs, is then compared against a signal line.</p><h4><strong>4. Signal Line</strong></h4><p>A signal line (usually a 9-day EMA of the MACD itself) is used to detect shifts in momentum.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BUhRb6_9SBgXCnsXNkI-Ew.png" /><figcaption>Sales and MACD for Trend Analysis</figcaption></figure><p>The interaction between the MACD line and the signal line provides crucial insights into momentum shifts. A <strong>bullish crossover</strong>, where the MACD line ascends above the signal line, typically suggests strengthening upward momentum, indicating a potential increase in the underlying data. Conversely, a <strong>bearish crossover</strong>, where the MACD line dips below the signal line, signals a potential decline in momentum.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Tzp6hqNUlKagQeCJu3D5qA.png" /><figcaption>Sales and MACD for Trend Analysis</figcaption></figure><h4><strong>5. Histogram</strong></h4><p>To further enhance the visualization and interpretation of these momentum shifts, a histogram is often employed. This histogram represents the difference between the MACD line and the signal line.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*B_guzY3HxLjp7DGdzxyuSw.png" /><figcaption>Growth Analysis</figcaption></figure><p>When the MACD line is above the signal line, the histogram bars will be positive, extending above the zero line, and will grow taller as the upward momentum strengthens.</p><p>Conversely, when the MACD line is below the signal line, the histogram bars will be negative, appearing below the zero line, and will grow deeper as downward momentum intensifies. The shrinking of the histogram bars (whether positive or negative) can also act as an early warning sign of a potential reversal in momentum.</p><h3>To Your Dashboard…</h3><p>While traditionally applied to financial analysis, the MACD’s versatility allows for its adaptation to various datasets. For instance, in e-commerce analytics, KPIs can be treated as the “price,” enabling the application of MACD to identify trends and shifts in KPIs momentum, providing valuable insights for business strategy-Like Sales, Sessions, Active Users, Conversion rates, and etc.</p><h4>1. Calculate Metric’s Lines — Sample KPI: Sales</h4><ol><li><strong>MACD Line:</strong></li></ol><ul><li>Calculate the 12-day Exponential Moving Average (EMA) of daily sales figures.</li><li>Calculate the 26-day EMA of daily sales figures.</li><li>Subtract the 26-day EMA of sales from the 12-day EMA of sales. This difference forms your “Sales MACD Line.”</li></ul><p><strong>2. Sales Signal Line:</strong></p><ul><li>Calculate the 9-day EMA of the “Sales MACD Line.” This is your “Sales Signal Line.”</li></ul><ol><li><strong>Sales Histogram:</strong></li></ol><ul><li>Subtract the “Sales Signal Line” from the “Sales MACD Line.” This forms your “Sales Histogram.”</li></ul><h4><strong>2. Interpreting MACD for Sales Analysis:</strong></h4><p><strong>Sales MACD Line crossing above Sales Signal Line (Bullish Sales Signal):</strong></p><ul><li>Indicates that short-term sales momentum is accelerating relative to longer-term momentum.</li><li>Suggests a potential increase in sales growth.</li><li>Could be a signal to increase marketing spend, prepare for higher demand, or stock up on inventory.</li></ul><p><strong>Sales MACD Line crossing below Sales Signal Line (Bearish Sales Signal):</strong></p><ul><li>Indicates that short-term sales momentum is decelerating relative to longer-term momentum.</li><li>Suggests a potential slowdown or decline in sales growth.</li><li>Could be a signal to re-evaluate marketing strategies, clear excess inventory, or analyze competitor activity.</li></ul><h4><strong>3. Sales Histogram:</strong></h4><ul><li><strong>Bars above the zero line and increasing:</strong> Growing positive sales momentum. Sales are accelerating strongly.</li><li><strong>Bars below the zero line and decreasing:</strong> Strengthening negative sales momentum. Sales are decelerating or declining at an increasing rate.</li><li><strong>Height of bars:</strong> Larger bars indicate a stronger divergence between the “Sales MACD Line” and “Sales Signal Line,” implying stronger momentum in either direction.</li></ul><h3><strong>Practical Application</strong></h3><ul><li><strong>Marketing Campaign Effectiveness:</strong> Observe how the MACD responds after launching a new marketing campaign. A strong bullish cross and rising histogram could indicate success.</li><li><strong>Seasonal Trends:</strong> Use MACD to identify shifts in momentum related to seasonal sales patterns.</li><li><strong>Promotional Impact:</strong> Analyze the MACD to gauge the immediate and sustained impact of promotions and discounts on sales momentum.</li><li><strong>Inventory Management:</strong> Anticipate increases or decreases in demand based on MACD signals to optimize inventory levels.</li></ul><h3>Final Thoughts…</h3><p>To use this indicator effectively, we need to define the time windows for the short-term and long-term moving averages. These choices should be guided by the <strong>business logic</strong> and how the concept of a “trend” is expected to behave within a specific context.</p><p>In some businesses, trends might emerge and shift over daily or weekly intervals; in others, changes may only be meaningful over monthly or yearly timescales.</p><p>Based on my own experience, I’ve found that for most <strong>B2C business models</strong>, key behavioral metrics respond well to a <strong>7-day vs. 30-day window</strong> setup.</p><p>That said, these settings, including the length and structure of the time windows, can and should be <strong>customized based on the specific nature of each business</strong> and the behavior of its key metrics.</p><h3>Reference &amp; Code</h3><p>If you’d like to explore the code behind this analysis, or adapt it for your own business KPIs, you can find the full implementation here:</p><p>👉 GitHub Repository (Python and SQL Version): <a href="https://github.com/mahdinasseri/kpi-trend-analyzer">MACD-Based Trend Detection for Business Metrics</a></p><p>This project was developed with the assistance of generative AI to combine technical analysis concepts with real-world business analytics.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b549e324958a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Paradox of Our Mind’s Averaging of Environmental Data — A TEDx Naqshejahan Experiment]]></title>
            <link>https://medium.com/@mahdinasseri/the-paradox-of-our-minds-averaging-of-environmental-data-a-tedx-naqshejahan-experiment-387d425518b3?source=rss-ee8cd6238135------2</link>
            <guid isPermaLink="false">https://medium.com/p/387d425518b3</guid>
            <category><![CDATA[tedx-talk]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[cognitive-psychology]]></category>
            <category><![CDATA[average]]></category>
            <category><![CDATA[human-behavior]]></category>
            <dc:creator><![CDATA[Mahdi Nasseri]]></dc:creator>
            <pubDate>Thu, 10 Jul 2025 18:41:25 GMT</pubDate>
            <atom:updated>2025-07-10T18:41:25.903Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>The Paradox of Our Mind’s Averaging of Environmental Data — A TEDx Naqshejahan Experiment</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eeM0ZwZtNqMv68BBI4jg3w.png" /><figcaption>image credit: faceresearch.org</figcaption></figure><p>Imagine someone hands you the registration data for about 300 attendees of an event and asks, ‘Tell us what interesting insights you can find in this.’ What would you do?</p><p>That was exactly the request I received from Mohammad Zahedi, the organizer of <a href="https://www.ted.com/tedx/events/11182">TEDx Naghshe-Jahan</a> (now rebranded as <a href="https://www.ted.com/tedx/events/56241">TEDx Esfahan</a>). It felt a bit like being given half a kilo of tomatoes and being told to produce a whole kilo of tomato paste — enough to cook dinner for 300 people at an extravagant party!</p><p>It certainly wasn’t a proposal I could easily turn down. So, for several weeks, day and night — at home, at work, lying in bed, even in the shower — my mind was constantly churning,<strong> searching for an idea worthy of presenting on the TEDx stage</strong>. I pored over the data countless times, hoping that if I looked hard enough, the data itself would speak to me. But with large datasets, it often takes some effort to extract meaningful and impactful stories.</p><p>You might find it cliché to hear, but it was actually in the shower one day that I had my ‘aha!’ moment. Over those three weeks, there was really just one thing I kept wrestling with:<strong> the average</strong>. I was averaging the participants’ ages, their gender distribution, registration times, fees paid, and so on. But that was pretty much it — the data didn’t offer much more.</p><p>The core idea that gradually took shape in my mind was to focus on the very concept of averaging, and <strong>to highlight the pitfalls this widely-used tool can create in our lives</strong>. The average is a statistical instrument that lets us condense a scattered set of data into a meaningful single value — a metric that supposedly represents them all. That’s the standard, almost instinctive definition of average, and it’s hard to find a facet of modern life where some form of averaging isn’t at play.</p><p>What’s particularly fascinating is that <strong>our brains, too, rely heavily on the concept of averages in many cognitive processes</strong>. From the way our visual system perceives colors (often sensing an average of hues) to how we remember people or events (storing and recalling an average impression rather than each detail), our minds are constantly summarizing. This is the very process I tried to explore in my talk, and to demonstrate just how misleading — even erroneous — it can sometimes be.</p><p>It was, in fact, my experience collaborating with various companies on data analysis that made me question my trust in this age-old, powerful tool. Averages come with a host of issues, which I touched on in my TEDx talk, aiming to shed light on the subtle but profound impact this <strong>mental shortcut</strong> has on how we experience and interpret life.</p><h3>Nobody Is Average — and the Average Is Nobody</h3><p>The central point of my talk (you can watch it on <a href="https://www.ted.com/talks/mahdi_nasseri_nobody_is_average_and_average_is_nobody">TED</a> or <a href="https://www.youtube.com/watch?v=vkMHnnJSuf8">YouTube</a> in Persian) was that ‘<strong>nobody is average and average is nobody.</strong>’ By taking averages — especially when it comes to qualitative, human-centric data — we end up hiding the richness of all those individual details, relying on a summary value that matches none of the real data points.</p><p><strong>This mental miscalculation can create cognitive and emotional tension.</strong> But why does this happen? Why does our brain, with its unique ability to average, sometimes blunder so dramatically? That’s the question I set out to answer, though I won’t rehash my entire talk here — I’d rather encourage you to watch it instead. However, I’d like to share an interesting experiment we conducted while preparing the talk, in collaboration with the TEDx organizers.</p><h3>Testing the ability of participants’ minds to average and discover a contradiction in it</h3><p>To back up the main idea of my presentation, I needed a real-life experiment with real people to show that, <strong>in some scenarios, our brains are actually quite poor at averaging. </strong>And what better community for this experiment than the TEDx attendees themselves? We designed a short form with four questions, and — thanks to the organizers — emailed it to all registered participants at TEDx Esfahan. Of course not everyone responded, but we had a statistically sufficient sample to draw some basic conclusions.</p><blockquote>My central hypothesis proposes a contradictory function of the brain when mentally averaging external data. It performs with high accuracy under one set of conditions, yet with significant inaccuracy under another.</blockquote><p>Our brains tend to be extremely precise at averaging quantitative data from our environment, but when it comes to qualitative questions, accuracy drops dramatically. To test this hypothesis, we developed two sets of questions. In the first set, people were asked to estimate what they thought the average weight and average happiness level were among other TEDx attendees. In the second set, participants reported their own actual data — so we could use both sets for comparison. The two data points were: the participants’ own weight (a clear, quantitative metric) and their self-reported satisfaction or happiness level (a more qualitative variable). The questions were:</p><h4>Set One: Guess the averages</h4><ul><li>On average, how many out of ten TEDx NaghsheJahan participants report feeling happy in their lives?</li><li>What do you estimate is the average weight (in kilograms) of TEDx Naghshe Jahan participants?</li></ul><h4>Set Two: Share your actual data</h4><ul><li>Do you feel happy in your life?</li><li>How much do you weigh?</li></ul><p>I hope the distinction between these questions is clear. The first asks people to estimate the average of a real dataset, while the second gathers the real numbers, allowing us to directly compare perceived versus actual averages and see how well people’s instincts line up with reality.</p><h3><strong>The results were pretty much as I expected</strong></h3><p><strong>Let’s look at the first question: people’s weight.</strong> The true average weight was 72.5 kg, while participants’ estimated average was 72.3 kg — an error margin of just 0.3%. That’s remarkably precise! Our brains are clearly capable of making very accurate predictions about well-understood, quantitative measures like weight, possibly because we have a shared, universal understanding of what weight means — and perhaps because our brains evolved to handle these kinds of tangible metrics.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*IWnpQxqzLGsrFnTJ0ofWVA.png" /><figcaption>Comparing the participants’ actual average weight with their average guess</figcaption></figure><p><strong>But when it came to the qualitative variable, the results were strikingly different. </strong>On average, 79% of participants said they felt happy in their lives, while their estimate of the happiness of others was just 52% — a staggering 34% gap. That might seem like a reasonable error for a guess, but compared to the accuracy with weight, it highlights an intriguing weakness in our brain’s ability to average qualitative information.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*4lMZOAC5_9zJo8fX5gR1Pw.png" /><figcaption>Comparing the participants’ average feelings of happiness with their average guesses</figcaption></figure><h3><strong>What does this simple experiment show? An unskilled brain in averaging</strong></h3><p><strong>First</strong>, our minds are equipped with a highly precise mathematical tool and often leverage averaging as part of their processing power.</p><p><strong>Second</strong>, while we’re highly accurate at averaging quantitative metrics like weight, we’re almost hopelessly off the mark with qualitative traits like happiness.</p><p>This report is a rundown of the experiment we designed as a proof point for my talk’s central idea. As for how this experiment relates to the concept that ‘nobody is average, and average is nobody,’ I invite you to watch the video for the full story.</p><p><a href="https://www.ted.com/talks/mahdi_nasseri_nobody_is_average_and_average_is_nobody">Nobody is average and average is nobody | Mahdi Nasseri | TEDxNaghsheJahan</a></p><p>You can view the presentation slides on <a href="https://docs.google.com/presentation/d/e/2PACX-1vThJjgmHy0geno79XlEvCL4g0SLG_SEE3VMI6A0l01Y_sev6VKyBKX7Lf5kU0D1PLv0ncrjraaXeX4X/pub?start=false&amp;loop=false&amp;delayms=300&amp;slide=id.g36d4533102e_1_0">Google Slides</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=387d425518b3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Data Is Power — But How Do You Distribute It Across Your Organization?]]></title>
            <link>https://blog.startupstash.com/data-is-power-but-how-do-you-distribute-it-across-your-organization-2284372e4f6a?source=rss-ee8cd6238135------2</link>
            <guid isPermaLink="false">https://medium.com/p/2284372e4f6a</guid>
            <category><![CDATA[data-analytics]]></category>
            <category><![CDATA[data-driven-organization]]></category>
            <category><![CDATA[organization-structure]]></category>
            <category><![CDATA[decision-making]]></category>
            <category><![CDATA[data]]></category>
            <dc:creator><![CDATA[Mahdi Nasseri]]></dc:creator>
            <pubDate>Fri, 04 Jul 2025 14:11:00 GMT</pubDate>
            <atom:updated>2025-07-24T04:15:55.861Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>Data Is Power — But How Do You Distribute It Across Your Organization?</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*dzgRPESLFYSG6B9w" /><figcaption>Photo by <a href="https://unsplash.com/@grstocks?utm_source=medium&amp;utm_medium=referral">GR Stocks</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>Not just a tool for exercising power, but power itself</h3><p>If you haven’t yet established a data team or unit in your company, you may already be a bit behind. Chances are you’re lagging, but it’s definitely ‘not too late’ to start. Otherwise, you’ve probably already wrestled with the critical challenge of <strong>embedding data into your organizational structure. </strong>Let me assure you:</p><blockquote>It doesn’t matter how much high-quality data your business possesses, until you successfully integrate <strong>data usage</strong> into your <strong>organization’s structure</strong>, you can’t leverage data as a strategic advantage for growth.</blockquote><p>I’ve worked at and consulted for a variety of organizations, both large and small. My experience has shown that <strong>there’s no one-size-fits-all solution</strong>. The question, How can we use data as a competitive advantage? is generally the first, perhaps the most important, question every founder or CEO must ask. Not only is there no single straightforward answer, but with the rising complexity of data-driven operations (especially with the recent growth of AI), answers become more nuanced day by day.</p><p>I firmly believe that <strong>data is power, </strong>not just a tool for exercising power, but power itself, because increasingly, it’s the data making the decisions. You might think, But it’s humans making decisions, not data, but if you truly believe that, chances are you haven’t passed through technology’s half-open doors toward new horizons, and you risk being left behind. I won’t dwell on this point any further here. Instead, I’d like to explore:</p><blockquote>if <strong>data is power,</strong> then how should it be embedded at the organizational level? What organizational structure ensures this power is properly distributed and available to everyone, so the business can grow and thrive for the long haul?</blockquote><p>What I’ve learned over the years is that <em>decision-making in an organization</em> goes beyond being an innocent and creative attempt at problem solving; decision-making defines how power operates in every organization. When we talk about organizational decision-making, we’re really talking about a hierarchy of delegated authority. So when we say ‘data is power,’ it logically follows that <strong>an organization’s data analytics structure operates best when it’s designed and implemented in alignment with those power structures</strong>. Any other approach, in my experience, just wastes organizational resources.</p><p>Given all this, what’s the best model for embedding analytics in an organization? In this article, I’ve tried to answer this question. My focus here is on structures related to how data drives decision-making. I won’t be covering areas like data infrastructure, BI platforms, AI-driven features (Data Science), or data products. <strong>My focus is on the analytics unit, specifically, the segment of data utilization that supports human decision-making.</strong></p><p>Like every organizational activity, data analytics has plenty of complications: team building and hiring, designing compensation and benefits, career development, conflict resolution, inter- and cross-team coordination, and more, all of which are fundamentally issues rooted in structure. It’s the efficiency and effectiveness of the structure that determines how those challenges will be managed and how much organizational ‘friction’ you’ll experience. So, designing an effective structure for analytics within both operational and strategic processes is the kind of foundational work that, if done poorly, can lead the whole organization astray.</p><p>In the rest of this article, I’ll share my own experiences and insights through the lens of a useful framework presented in the article <strong>“</strong><a href="https://www.altexsoft.com/blog/how-to-structure-data-science-team-key-models-and-roles/"><strong>How to Structure a Data Science Team</strong></a><strong>”</strong>, published on the AltexSoft technology consulting blog. I highly recommend reading the original article as well for added context. Also, bear in mind:</p><blockquote>The following models share many features; some could even be considered counterparts. Use them as templates and ask yourself how analytics is currently embedded at your own organization, and whether this structure is truly the best fit for you.</blockquote><h3>How to embed a data analytics team in your company?</h3><h4>1. Decentralized Model</h4><p>This model is the most common pattern for the organic, gradual formation of an analytics unit during a company’s early stages. Business units, like product teams or functional groups, recognize their analytics needs as their domains evolve and begin to hire data specialists to fulfill these needs within their own teams.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/990/1*HRjgF4H0chDbrvVhtnYdUg.png" /></figure><p>This process requires <strong>minimal inter-organizational coordination</strong>: every team is responsible for meeting its own data needs, and each manager hires data talent as required. Sometimes, a current team member will gradually take on analytics tasks without any hiring at all. For example, a backend developer might eventually handle product analytics and become the team’s main resource for data-driven product decisions.</p><p>This model fits two scenarios. First: <strong>companies still in their growth phase,</strong> without a traditional organizational structure. I believe this is actually the best approach for any startup, let your analytics function grow organically alongside the organization. The crucial advantage here is that the <strong>structure emerges from actual needs</strong>, making it easier to control costs.</p><p>Second: <strong>organizations that aren’t in early growth</strong>, but whose business model is never going to be truly <em>data-centric</em>. That is, data isn’t genuinely a <em>capital asset</em> for them. This is rare among tech companies, but common in physical businesses, many of which may never need data beyond routine reporting. These companies are focused on operational efficiency rather than optimizing decision-making. In such contexts, it’s better to let every business or functional unit hire data talent only as needed.</p><p>In any case, this model, like any, has its weaknesses, especially if applied in contexts outside those described. It often leads to siloed departments, lack of standardized analysis, and ultimately decentralized reporting. Data insights simply don’t travel well between units, and each group ends up with its own data silo. This becomes critical as data volumes grow and the absence of a centralized data analytics infrastructure at the company level results in significant wasted resources.</p><p>The next challenge: <strong>hiring itself</strong>. Department managers often lack the skill set to recruit a data analyst. For example, a software engineering manager understands engineering talent inside and out, but may be unfamiliar with the required roles and abilities of a data analyst, which leads to hiring headaches.</p><p><strong>Retaining analytics talent is also a problem here. </strong>While a software manager knows how to help an engineer advance, designing and developing a career path for a data analyst within a single, isolated team becomes problematic, especially in the long run. Few data analysts can work for over a year in such an isolated setting.</p><p>In my opinion, the <strong>biggest issue with this model is the lack of growth in the quality standards of data analytics tasks</strong>. Experience shows that analysts in this structure are the least likely to follow best practices, adhere to standards, or experiment with different techniques. This is where structure can misguide people, every data analyst needs to work with more senior and experienced colleagues to learn from them and stay aware of different problem-solving approaches. Here, an analyst is left to fend for themselves, if they’re not intrinsically motivated to learn and grow, the system does little to help them improve or upgrade their work.</p><h4>2. Functional Model</h4><p>In this model, a team of one or more data analysts forms within the most relevant functional area, usually marketing or product, for example, where analytics needs are the greatest and, over time, a cluster of analytics talent, tools, and domain knowledge emerges.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/990/1*B8aB1uk6tRv9KLSPdcOwSA.png" /></figure><p>Like the decentralized model, the functional model <strong>tends to evolve organically and gradually along with the organizational structure</strong>. Similarly, it reaps minimal benefits from coordinated, company-wide analytics at both operational and strategic levels.</p><p>That said, this model can work as a <strong>first step toward systematic organization-wide analytics</strong>. In other words, if your company hasn’t had a dedicated analytics unit (or only a decentralized one), you could set a mid-term goal of establishing a functional analytics model within your structure.</p><p>In this approach, start with your most relevant business unit. Build an analytics team there, and help it grow in a systematic way, developing relevant skills so it becomes a center of knowledge for building analytics capabilities elsewhere in the organization.</p><p>The analytics team you establish (say, in marketing or product) could handle requests from other business or functional units for some time, and thus, you can develop a specialized analytics team within part of your organization at minimal cost, benefiting from its expertise.</p><p>The <strong>downsides of the functional</strong> model lie in its very centralization. The first problem (like the decentralized approach) is disengagement from broader company issues. Here, analytics efforts focus solely on the needs of the team they’re embedded in, not the organization as a whole. That lack of perspective can isolate analytics, and even cause misalignment with the company’s strategy.</p><p>The second issue is <strong>weak integration of analytics for strategic growth — largely due to the absence of a company-wide data lead</strong>. In this model, the analytics team falls under a specific business unit, reporting directly to its head, so data priorities are shaped by that team’s needs. The analytics function ends up boosting only local capabilities, not the entire company’s. And while teams usually claim to be aligned with organizational strategy, experience has shown me that, at any level of cultural maturity, teams tend unconsciously to reinforce their own position, which doesn’t always benefit the larger organization.</p><p>Therefore, organizations looking to use data as a company-wide enabler should <strong>never allow <em>data power</em> to remain the exclusive domain of a single team</strong>. That’s the pitfall of the functional model.</p><h4>3. Consulting Model</h4><p>In this structure, analytics specialists work together as one team but operate in a consultative capacity for the rest of the organization: <strong>different departments ‘borrow’ them for specific tasks.</strong> Structurally, this advisory model is similar to a centralized one (which we’ll consider next), but here, the analytics team isn’t solely responsible for execution, each team is responsible for its own analytics work, ‘borrowing’ consultants from the analytics team as needed. Once the project is done, consultants are reallocated elsewhere, acting almost like seasonal contractors moving from issue to issue.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/990/1*BSmguIjvQdrS9YGOgSILPQ.png" /></figure><p>In this model, the analytics team’s main role is <strong>talent management</strong>: allocating resources among teams, prioritizing cross-team requests, and hiring analytics professionals.</p><p>The consultative model is well-suited for small and medium-sized businesses with sporadic or moderately sized analytics projects. Since every team member reports to a central analytics manager, managing such a team is more straightforward and affordable for these organizations.</p><p>Nevertheless, this approach carries its own weaknesses and doesn’t fit every structure. Chief among them: <strong>it’s not conducive to analytics quality</strong>. Temporary assignment to teams prevents analysts from gaining deep domain knowledge, so they can’t always implement best practices. Lacking time to learn the finer points, and given that other organizational priorities often emerge, analysts are tempted (or forced) to favor quantity over quality and chase quick, superficial solutions.</p><p>This model is also weak when it comes to creating and sustaining <strong>motivation</strong>. When a data analyst isn’t fully engaged in product development or strategic decisions, it’s hard to foster a sense of ownership or to stay motivated for great results.</p><p>In my experience, every analyst here eventually ends up juggling projects from multiple teams. Managing conflicting priorities and deadlines across departments becomes a major headache for analytics teams, often resulting in burnout and eroding confidence in project outcomes.</p><h4>4. Centralized Model</h4><p>The <strong>biggest issue</strong> with all three models so far is this: <strong>scattering the analytics function across business units means that cross-company strategic questions often go unanswered.</strong> A centralized analytics team, on the other hand, is much better positioned to tackle broader organizational challenges and thus extract maximum value from data, a team that, aligned with leadership’s strategic priorities (and not limited to just a handful of departments), addresses key projects in a data-driven way, while still acting as a contractor for day-to-day analytics tasks from other teams.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/990/1*h9B5kyrIhpmi8x7tsCFNzg.png" /></figure><p>The close coordination and strong shared purpose of analytics specialists in a centralized team, guided by a single manager, leads to higher <strong>productivity</strong>; deeper teamwork and <strong>knowledge sharing</strong>; faster professional development and clearer <strong>career paths</strong> for data professionals; and ultimately, if things go well, the organization ends up with a powerful analytics group that can focus on <strong>high-priority, strategic issues </strong>without being mired in operational distractions.</p><p>One of the best moments to form a centralized analytics team is during <strong>rapid company growth, </strong>when demand for analytics suddenly spikes and you urgently need analytical expertise. At such times, setting up a centralized team and meeting all analytics needs via a single gateway is the most efficient approach. Failing that, your organization’s data-driven decision making will fall behind as your organizational structure evolves.</p><p>A centralized analytics team also sends a strong signal company-wide: <strong>that leadership intends to use data as a strategic competitive advantage.</strong> Since the analytics team lead reports directly to the CEO, no other department can simply ignore it, which helps ensure your fast-evolving company culture matures around data.</p><p>Of course, the centralized model, like any other, has its cons; it’s arguably the most <strong>risk-averse</strong>. Whenever I see a company maintain this model for a long time, I can only conclude they’re reluctant to share business data across the organization, even with their own employees. In such a culture, data access is restricted to a chosen few for tight control, which makes it tough for people who want to work data-driven on their own responsibilities.</p><p>Furthermore, in a structurally mature company, it’s quite likely that the centralized analytics team will eventually become <strong>isolated</strong>, retiring from active engagement with other departments. Without direct involvement in operational and product activities, data gradually gets sidelined from key business decisions. Over time, the analytics team loses direct understanding of the company’s true value drivers, and their recommendations, or data-driven solutions, drift further from real-world problems. <strong>Other teams may then ignore or dismiss those recommendations.</strong></p><p>This creates challenges for meaningful collaboration with product teams. When the analytics team presents a solution, it might not fit the product roadmap, leading to conflict. Conversely, if the product team has an analytics request, working with the analytics team, defining requirements, clarifying a data-driven path, and completing the work, can require so much time and overhead that others eventually give up on seeking analytics support altogether.</p><p>While a centralized model might seem to be the simplest, most effective structural option for analytics, I have yet to see a mature organization sustain this approach for more than about six months. Over time,<strong> the analytics team loses touch with business realities</strong> and, at best, morphs into a passive unit, dealing only with regular reporting and maintaining data access: a kind of professional purgatory for any analyst.</p><p>At this point, it’s clear that analytics can only serve as a business enabler, and convert data assets into a competitive edge, <strong>when it flows, like lifeblood, through every part of the organization</strong>. So we need a model that enables analytics to inform day-to-day decisions everywhere, but that can also coordinate at a strategic, company-wide level. The next two models offer potential solutions.</p><h4>5. Center of Excellence Model</h4><p>One big issue with decentralized models was the difficulty of developing human capital and the tendency for knowledge to get siloed by unit. One way to address this is to create a Center of Excellence at the organizational level.</p><p>A Center of Excellence (CoE) is a group of employees with a shared focus and deep subject expertise. Its purpose is generally to provide guidance, insights, training, and skill development across the organization. A <strong>data analytics CoE</strong> is an independent unit, separate from other teams, providing supportive help, rather than direct execution, to analysts across the company.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/990/1*bx96I94auZFLgFmg4KmpNg.png" /></figure><p>In this setup, the CoE isn’t accountable for delivering actual analytics projects; it exists to <strong>support, enable, and coordinate</strong> cross-team analytics work. Tasks like hiring, upskilling analysts, promoting collaboration, providing infrastructure tools, and helping plan career development all fall under its remit. Actual analytics work still belongs to each business or functional team’s in-house analysts.</p><p>Here, a unified coordination center is preserved, but analysts are embedded within various departments. The structure is a hybrid of centralized and decentralized models, analytical activities are strategically coordinated top-down, but specialists remain closely involved with business operations.</p><p>Because this approach balances the advantages and drawbacks of centralization and decentralization, it’s increasingly popular, especially in large organizations. It works best for firms that have mature data strategies and roadmaps.</p><p>Still, even a coordination-heavy model like this has its flaws. There’s no dedicated, centralized team focused solely on company-wide strategic analytics; each analytics group solves only its own functional problems. The Excellence Center may prevent redundant work, miscommunication, and skills gaps, but under heavy workload, some analytics still happens in isolated ‘islands,’ and some decisions may not truly serve the whole business.</p><p>In my view, the biggest gap in this model is the<strong> lack of any truly data-driven innovation unit</strong>: a group of data specialists actively exploring new solutions and pursuing long-term, groundbreaking data projects beyond routine needs.</p><h4>6. Federated Model</h4><p>To truly improve and elevate the CoE model, some form of centralized analytics team is still needed. The ‘federal’ model gets close, a hybrid familiar from federal governments. Here, the organization has a central analytics team that manages and coordinates all analytics teams embedded in different departments (like independent states). But the central team’s role isn’t limited to CoE-style coordination; it includes its own data analysts who can independently, under the direction of the central lead, <strong>tackle strategic or cross-functional projects as needed</strong>. This lead reports straight to the CEO and works closely with business and functional units.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/991/1*4EzCY9mkFCE-gj5HVMOGfw.png" /></figure><p>In this model, each department can have its own analytics unit, while the whole organization benefits from a central team for organization-wide analytics projects. The central federal team manages<strong> cross-department coordination, human capital, knowledge sharing — and most importantly, project quality — across the distributed analytics teams</strong>.</p><p>Compared to the other models presented so far, this one is structurally more complex, a level of complexity only suitable for large organizations willing to bear the associated costs. It’s usually not the best fit for companies in the earliest stages of analytics adoption.</p><p>Finally, the <strong>multi-functional nature</strong> of the federal team may create organizational tension. Because of its strategic mandate, the federal team can involve itself in any department’s domain, take on its own projects, and pursue operational goals in parallel with other teams. This can generate power struggles among team leads and upset organizational balance.</p><h4>7. Democratic Model</h4><p>This isn’t necessarily a standalone model, it can be combined with other approaches. The democratic model gives everyone in the organization access to data via BI tools or data portals. You might blend a federal model with a CoE and analytics talent embedded everywhere — while also letting anyone who wants to use data access what they need through BI self-service.</p><p>My experience shows that this democratized approach strongly encourages and enhances a<strong> data-driven culture</strong>. Here, product team members, product managers, engineers, designers, developere, as well as business, marketing, and other managers can access data directly, without needing a dedicated analyst.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/990/1*v4GGqMJMXTqZG1pVEgyOJg.png" /></figure><p>I always recommend <strong>democratizing data access</strong>.</p><blockquote><strong>If we accept that ‘data is power,’ and that ‘decision-making equals exercising that power,’ the logical conclusion is that all organization members should have democratic access to data — so anyone can use it to make better decisions.</strong></blockquote><p>It’s crystal clear to me that if data access is monopolized by any group, the organization will quickly see <strong>corruption in decision-making</strong> at the operational and strategic level.</p><p>But what are this model’s weaknesses? <strong>Democracy isn’t cheap</strong>. Companies that implement this model typically invest heavily in data infrastructure, tools, and training. You’ll need an independent BI team to keep these tools current and efficient. At the same time, offering a universal BI solution to everyone tends to reduce local effectiveness over time, and may eventually result in those tools being discarded.</p><h3>Final Thoughts…</h3><p>Organizations are living systems, and as they grow, their structure must evolve, too. Remember: <strong>the analytics model that works for you today may need to change as the business changes.</strong> Be prepared to invest the necessary time, costs, and experiments, so that you eventually land on the right model for embedding analytics into your processes, and set yourself up for strategic growth in the long run.</p><p>I have previously published this article in Persian at this <a href="https://vrgl.ir/ULKSB">link</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2284372e4f6a" width="1" height="1" alt=""><hr><p><a href="https://blog.startupstash.com/data-is-power-but-how-do-you-distribute-it-across-your-organization-2284372e4f6a">Data Is Power — But How Do You Distribute It Across Your Organization?</a> was originally published in <a href="https://blog.startupstash.com/">Startup Stash</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Unmarketing, Unproduct, Uncustomer]]></title>
            <link>https://blog.startupstash.com/unmarketing-unproduct-uncustomer-cd5ea74f6f47?source=rss-ee8cd6238135------2</link>
            <guid isPermaLink="false">https://medium.com/p/cd5ea74f6f47</guid>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[analytics]]></category>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[marketing-strategies]]></category>
            <dc:creator><![CDATA[Mahdi Nasseri]]></dc:creator>
            <pubDate>Thu, 12 Jun 2025 12:30:43 GMT</pubDate>
            <atom:updated>2025-06-16T11:10:40.588Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BhBwoUdNF7xNpJXhN1FReA.jpeg" /><figcaption>Photo by Artem Beliaikin @belart84 on Unsplash</figcaption></figure><h3><strong>1. What a Young Entrepreneur Sees in the Mirror, Peter Drucker Sees in Plain Clay</strong></h3><p>You’ve probably heard this famous quote from Peter Drucker: <em>“The best way to predict the future is to create it.”</em></p><p>Many consider it his most well-known saying. But I believe Drucker has said far more important and useful things — <strong>especially in the field of marketing and strategy.</strong></p><p>The reason I’m writing this post is one of those powerful insights Drucker shared about marketing. Let’s read it carefully:</p><blockquote><em>“Everyone knows that marketing and sales are not the same. But few understand that they are not necessarily complementary. In fact, they can even be in conflict.”</em></blockquote><blockquote><em>~ Peter Drucker</em></blockquote><p>It sounds obvious, right? But I suggest reading it again — and asking yourself honestly: <strong>how much do you really agree with this in practice?</strong></p><p>Can you name examples in your business where sales and marketing have clashed? Why do so many managers and marketing teams, often unconsciously, treat marketing as just sales with a bit of polish?</p><p>How is it possible that successful sales can actually work against marketing? And in such cases, who’s right — sales or marketing?</p><p>And most importantly, <strong>why can ignoring this fundamental difference put your business at risk?</strong></p><p>A while ago, during a nature tour, I asked a more experienced fellow traveler to recommend a good store for buying outdoor gear. He pointed me to a small shop and explained his reason like this: “<strong>The owner isn’t just a salesperson</strong> — he helps you choose the right product based on what you actually need.” Then he added, “For example, if you need a tent for a one-night picnic, he won’t let you buy a high-altitude mountaineering tent — even if you can afford it and insist on getting it.”</p><p>Conversations like this are probably familiar to many of us. We often hear similar stories when friends or family recommend a business or shop.</p><p>But what makes us recommend one business to others — and return to it ourselves — while ignoring another? My friend’s answer may sound simple at first, but it’s based on an important principle in marketing strategy.</p><p>His answer clearly highlights a common <strong>misunderstanding</strong> between two concepts: sales and marketing.</p><p>Understanding this difference is key. If we get it right, we’ll likely be able to build more effective marketing strategies and grow our startup or business in a sustainable way. But if we don’t, marketing will just become a costly effort with little value.</p><p>Let’s explore this idea further.</p><h3><strong>2. “Every Business Has Only One Purpose: To Create a Customer”</strong></h3><p>The next time you meet a business manager, marketing lead, product manager, or even someone from the product or tech team, ask them this question: <strong>What is the main goal of a business?</strong></p><p>In nearly every case, the answer will be: <strong>profitability. </strong>And if you hear something else, it’s probably because they felt awkward giving such an obvious answer to such a big question — and tried to reframe “profit” from another angle.</p><p>Now, let’s turn to another famous quote from Peter Drucker:</p><blockquote><em>“The purpose of a business is to create a customer.”</em></blockquote><blockquote><em>~ Peter Drucker</em></blockquote><p>Ask that same person a follow-up question: <strong>What is the most important task a business should focus on to reach its goal?</strong></p><p>They’ll likely say: <strong>sales.</strong></p><p>Now let’s look at Drucker’s full quote:</p><blockquote><em>“The purpose of a business is to create a customer. Therefore, a business enterprise has</em><strong><em> two — and only two </em></strong><em>— basic functions: marketing and innovation.”</em></blockquote><blockquote><em>~ Peter Drucker</em></blockquote><p>Drucker not only believed that marketing and sales are not the same — he often emphasized that in some cases, they are actually at odds with each other.</p><p>And it’s clear: he believed, as do we, that when there’s a conflict, <strong>marketing is usually right</strong>.</p><blockquote><em>“Sales focuses on the needs of the seller and the need to turn products into cash, whereas marketing focuses on the needs of the buyer and the need to satisfy the customer through the product.” <br>~ </em><a href="https://www.amazon.com/Marketing-Myopia-Harvard-Business-Classics/dp/1422126013"><em>Theodore Levitt, </em>Marketing Myopia</a><em>, explaining Drucker’s view on the difference between sales and marketing</em></blockquote><p>The first thing that comes to mind when reading quotes like these is this:<br><em>No matter what “creating a customer” means, it ultimately has to lead to maximizing profit — otherwise the business loses its purpose.</em></p><p>But Drucker strongly disagreed.</p><p>After studying dozens of successful businesses and advising some of America’s largest companies, he reached a different conclusion: <strong>The businesses that survived for decades were the ones that saw profit as a tool for creating customers — not the other way around.</strong></p><p>As Drucker put it:</p><blockquote><em>“Profitability is essential — but maximizing profitability is not.”<br> ~ Peter Drucker</em></blockquote><p>Profit is like oxygen for a business. We breathe to live — we don’t live to maximize the amount of oxygen in our blood. In fact, trying to do that could harm us.</p><p>Similarly, Drucker believed that maximizing profit brings serious downsides for any business.</p><ul><li>First, it <strong>poisons</strong> the organizational culture.</li><li>Second, high profit margins make your business look greedy — or even <strong>unethical</strong>.</li><li>Third, high profits create a false sense of success, leading managers to make poor strategic decisions based on <strong>short-term wins</strong>.</li><li>And fourth, high margins make it easier for competitors to enter the market and take market share. As a result, your business — which has been operating efficiently only under high-margin conditions — starts to lose its edge when faced with <strong>real competition</strong>.</li></ul><p>But there’s still one unanswered question: <strong>What does “creating a customer” actually mean?</strong></p><p>Understanding the difference between marketing and sales helps clarify this. It shows why <strong>marketing is what creates a customer</strong>, while <strong>sales is what creates revenue</strong>.</p><p>To better understand this difference, let’s introduce a new concept: <strong>Unmarketing</strong> — similar to how uses the term <em>Unhappy</em>.</p><p>There are two surprising things about Unsmarketing: First, most marketing teams are actually doing Unsmarketing rather than marketing. And second, Unsmarketing is often accompanied by <strong>successful sales</strong>.</p><h3>3. Unmarketing: What Marketing Is Not</h3><p>Unmarketing is everything that marketing <strong>should not be</strong>. It not only fails to create customers — it can even destroy existing ones or prevent new customers from being created. At best, Unsmarketing is a waste of energy and resources.</p><p>The harsh truth about Unsmarketing is this: <strong>It often overlaps with successful sales.</strong></p><p>In other words, there are many cases where sales happen, but the marketing behind them was wrong — or harmful. That’s when we call it Unmarketing.</p><p>To go further, let’s define a few more terms that help clarify the idea:</p><p><strong>Unproduct<br></strong>The opposite of a product — a solution that doesn’t match the customer’s needs, or doesn’t solve the problem properly. It doesn’t mean the product doesn’t “work” — for example, using a numbing drug to solve thirst technically “works” (you won’t feel thirsty), but it’s a terrible solution. A real product fits the problem. That’s <em>product-problem fit</em>.</p><p><strong>Uncustomer<br></strong>The opposite of a customer — someone who doesn’t fit your target market. Like trying to sell baby diapers to a single adult with no children. Anyone outside your target persona is a Unscustomer: they don’t really need what you’re offering, or your product isn’t built for them.</p><p><strong>Unsale<br></strong>The opposite of a sale — but intentionally. This isn’t a failed sale. It’s a <em>strategic decision</em> not to sell, even when you could. For example, choosing not to advertise on a platform where customers often expect heavy post-sale support — because you know the cost of supporting those customers would be too high.</p><p>This mindset should shape your entire marketing approach: <em>you actively choose who not to sell to</em>.</p><p>Every marketing activity results in a combination of these three dimensions:</p><ol><li>Sale or Unsale</li><li>Product or Unproduct</li><li>Customer or Uncustomer</li></ol><p>That gives us <strong>eight possible outcomes</strong>, which we’ll look at in the following chart.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bzZF1PbP30KL4NBtUfN_Ww.png" /></figure><p>Each path from the root to the leaves of this graph represents one possible outcome of a marketing or unmarketing scenario.</p><p>Let’s look at two specific examples, both labeled in the graph:</p><ul><li><strong>Path 1: Sale — Product — Customer<br></strong>This is a successful sale of the <em>right product</em> to the <em>right customer</em>. It’s what we aim for in ideal marketing: matching real customer needs with a proper solution and closing the sale.</li><li><strong>Path 2: Unsale — Product — Uncustomer<br></strong>This means <em>refusing</em> to sell (Unsale) a good product to the <em>wrong customer</em>. It’s a conscious decision not to proceed with the sale, even though the product is right — because the buyer is not part of your target audience.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*X0q8WuWmfwTG_o5BtDxO5w.png" /></figure><p>Now, with the help of this graph, we can clearly see which scenarios represent <strong>real marketing</strong> and which ones are actually <strong>unmarketing</strong> — that is, not marketing at all.</p><p>Out of the 8 possible combinations, <strong>4 represent proper marketing activities</strong>, while the <strong>other 4 fall under unmarketing</strong>.</p><p>Let’s say you’re the founder and CEO of a new startup called <strong>Cartoonino</strong> — an online video streaming (VOD) service focused exclusively on cartoons and animated films for children. Your goal is to fully capture the children’s segment of the VOD market. Your strategy is to run a <strong>guerrilla campaign</strong> against market leaders. You’re aiming to claim a small, well-defined niche and protect it so effectively that the big players can’t easily take it from you.</p><p>In the following section, we’ll explore what marketing and unmarketing look like through examples of real decision-making situations you might face as CEO of Cartoonino.</p><h3>3.1. What Is Marketing? <strong>Sell, Unsell, Unsell, Unsell</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Py93_gyrfpJ1FiHw3bIWhA.png" /><figcaption>Marketing: Sell, Unsell, Unsell, Unsell</figcaption></figure><h4><strong>1. Selling the right product to the right customer</strong></h4><p>A marketer should only aim for a successful sale when the <strong>right product</strong> is being sold to the <strong>right customer</strong>. That’s why marketing doesn’t start with advertising or sales — it begins much earlier, at the product design stage. It continues through customer research, market analysis, customer development, segmentation, and more.</p><p>In the startup world, this set of responsibilities is known as achieving <a href="https://leansteps.wordpress.com/11-2/step-3-lean-experiment/problem-solution-fit/"><strong>problem–solution fit</strong></a> and <a href="https://en.wikipedia.org/wiki/Product-market_fit"><strong>product–market fit</strong></a>.</p><blockquote><strong>In Cartonino…</strong><br> The <em>right product</em> is one that’s been designed and developed with both the <strong>content</strong> (cartoons and movies) and the <strong>user experience (UX)</strong> perfectly tailored to children between the ages of 5 to 13. Therefore, you strongly advise the UX team <strong>not to copy</strong> the user experience of leading VOD platforms, and instead, you ask them to design the entire user journey using <strong>A/B testing techniques</strong>. You even invite a few children to join the UX design team regularly and test the product throughout development.</blockquote><blockquote>On the other hand, your <strong>customer</strong> (not the end user) is the young parent — likely between the ages of 25 to 40, and financially in the <strong>upper middle class</strong>. Correctly identifying the customer (the parent, not the child) and designing all marketing strategies around <strong>attracting and retaining them</strong> is the first and most important marketing decision you’ll have to make. The tone of your messaging, advertising channels, and promotional offers should all be crafted specifically for this <strong>target customer segment</strong>.</blockquote><blockquote>If you succeed in selling the right product to the right customer, you’ve executed <strong>true marketing</strong>.</blockquote><h4><strong>2. Unselling the Product to the Uncustomer</strong></h4><p>Selling your product to someone who isn’t the <em>real customer</em> (uncustomer) for it may generate short-term revenue, but it has consequences that contradict the fundamental principle of <em>customer creation</em>. The un<em>customer</em> will eventually become dissatisfied with your product — not because of the product itself, but because your otherwise well-designed solution simply doesn’t match their needs.</p><p>This type of customer has a <strong>very low lifetime value (LTV)</strong> and often won’t go beyond the initial purchase. That means the <strong>customer acquisition cost (CAC)</strong> you spent is effectively lost and won’t be recovered — especially in <strong>B2B businesses</strong>, where the cost of acquisition and onboarding is usually high, and profitability relies heavily on the customer’s lifetime.</p><p>Avoiding this kind of sale is a <strong>strategic unsale</strong> — a conscious refusal to market to or serve someone who doesn’t truly fit your value proposition.</p><blockquote><strong>In the case of Cartonino…</strong></blockquote><blockquote>Your product has been developed over the past few months, passed all analytical and usability tests, and you’re now confident that Cartonino has an effective UX. You’ve also secured strong licensing agreements with producers of children’s content and gained access to a valuable archive of classic and international cartoons. The product is ready in every sense.</blockquote><blockquote>A friend of yours, who runs a successful nature-travel social media platform with hundreds of thousands of active users, offers to promote Cartonino to his audience at a very low cost. You consider it — but despite having a solid product, you decide to <strong>decline</strong> his generous offer. Why?</blockquote><blockquote>Because the users of this nature-travel platform <strong>do not belong to your target customer segment</strong>. They are mostly between 18 to 30 years old and primarily single. These users are <strong>uncustomers</strong> for your business. Advertising to them would result in nothing but <strong>wasted marketing spend</strong> and would likely <strong>weaken your brand positioning</strong>.</blockquote><h4><strong>3 — Unselling Unp<em>roduct</em> to Unc<em>ustomer</em></strong></h4><p>Everything that applies to the second scenario (not selling a good product to a non-customer) holds here as well — but it gets even worse. Because in this case, not only is there a mismatch between the customer’s needs and the product, but the product itself doesn’t even properly deliver on <strong>its own promise</strong>.</p><p>In other words, you’re dealing with a <strong>unproduct</strong> — a solution that either addresses the wrong problem or fails to do what it claims. Selling it to a <strong>uncustomer</strong> just amplifies the damage: poor product–customer fit combined with poor product integrity.</p><blockquote><strong>In Cartonino…</strong><br>Your marketing team informs you that the <em>Anime</em> section — available for free streaming — has been attracting significant traffic from Google. They suggest doubling down on this audience and even propose turning this category into premium content, requiring users to subscribe to watch. It’s an exciting analytical insight, and the team is quite enthusiastic about it.</blockquote><blockquote>You, however, decide to dig deeper into this trend before acting. After analyzing the data in Google Analytics, you discover several key points:</blockquote><blockquote>The demographics of these anime viewers — those landing on the site via Google — differ significantly from your defined customer persona.</blockquote><blockquote>Their age range is outside your target segment, and their search queries mostly include the term “free.” In short, they are <em>non-customers</em>.</blockquote><blockquote>Additionally, their <strong>time on page</strong> is significantly low.</blockquote><blockquote>You assign your tech team to investigate further and discover that your video streaming module has a bug — it fails to play some videos properly at specific resolutions, causing them to stop after a few minutes. This makes those anime videos a <strong>unproduct</strong> in practice.</blockquote><blockquote>Your final decision:<br>Rather than chasing short-term revenue from random anime seekers, you choose to refocus SEO efforts on high-intent search terms — like names of trending children’s shows and prioritize fixing the streaming issue.</blockquote><blockquote>This is a case of <strong>unselling </strong>an<strong> unproduct </strong>to an<strong> uncustomer</strong> — a conscious un<em>sale</em> that protects your brand, your strategy, and your resources.</blockquote><h4><strong>4- Unselling Unproduct to Customer</strong></h4><p>This is one of the <strong>toughest strategic decisions</strong> in marketing. Startups that experience <strong>sudden, rapid growth</strong> often face this dilemma. Demand surges before the product has become stable, refined, or fully functional. In these situations, it’s hard to rein in the temptation to scale aggressively, and at the same time, it’s difficult to enhance and stabilize the product fast enough to meet that demand.</p><p>Unless specific conditions — such as competitive pressure or market entry strategy — dictate otherwise, <strong>marketers should refrain from selling a unproduct to customers</strong>. Doing so risks disappointing early adopters, undermining brand trust, and creating churn.</p><p>In such a scenario, the marketing team’s role isn’t to drive sales, but rather to:</p><ul><li><strong>Accelerate the feedback loop</strong> between users and the product team</li><li><strong>Help define a lean product design process</strong></li><li>Ensure that the product is iterated quickly and meaningfully toward a true <strong>product-market fit</strong></li></ul><p>Selling prematurely — even to the right customers — can damage long-term growth and sabotage brand equity. Marketing, at its core, should be committed to <strong>creating customers</strong>, not just transactions.</p><blockquote><strong>In Cartonino…</strong><br> Your new investor insists on immediately launching an <strong>ATL (Above The Line)</strong> campaign using <strong>billboard advertising</strong>. Since major players have long been active in these physical advertising channels, the investor anticipates strong growth for Cartonino and is even willing to provide additional funding to support the campaign.</blockquote><blockquote>However, you know that the <strong>Cartonino app still lacks a video player properly tailored for children</strong> — the current player has a generic UX that, according to your research, <strong>kids struggle with significantly</strong> (unproduct).</blockquote><blockquote>You decide to have a <strong>strategic discussion with your investor</strong>. Instead of jumping into billboard ads, you propose reallocating part of the budget toward <strong>accelerating development of the video player UX</strong>, specifically optimized for children. You also recommend <strong>delaying ATL advertising</strong> for now.</blockquote><blockquote>Why? Because you don’t want to <strong>sell a unproduct to customers</strong>. Doing so could backfire, eroding trust and harming your brand in the long run. If early adopters encounter a frustrating experience after seeing polished, promising ads, they may not believe future versions will improve. In your view, <strong>earning and maintaining trust is more valuable than short-term growth spikes</strong>.</blockquote><p>The key insight here is that among the <strong>four valid marketing scenarios</strong>, <strong>only one involves making a sale</strong>, while the other <strong>three involve <em>not</em> selling (unsale)</strong>.<br> This is exactly where <strong>Peter Drucker</strong> famously argued that <strong>marketing and sales are, in most cases, fundamentally at odds</strong> with one another.</p><p>According to Drucker, the <strong>essence of marketing</strong> is to <strong>know the customer so well</strong> and understand their needs so deeply that the product or service is designed <em>based on that understanding</em> — so aligned with the customer’s needs — that <strong>the product essentially sells itself</strong>.<br>In this view, marketing is not primarily about persuasion or promotion — <strong>it’s about fit</strong>. The better the fit, the <strong>less selling is needed</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jqxLtrz1eas4CL2GcF1otA.png" /></figure><h3>3.2. What Is Unmarketing? <strong>Sell, Sell, Sell, Unsell</strong></h3><p>In contrast to the legitimate marketing cases outlined earlier, un<em>marketing</em> is essentially made up of the following four scenarios:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*br02A4sPl_dLG7PGyp1CBA.png" /><figcaption><strong>Unmarketing: Sell, Sell, Sell, Unsell</strong></figcaption></figure><h4>1. Selling the Product to the Uncustomer</h4><p>This is when a fundamentally good product is sold to someone who is not the right customer. It may generate short-term revenue, but comes with serious long-term costs:</p><ul><li><strong>High customer acquisition cost (CAC)</strong> for a customer with a <strong>low lifetime value (LTV)</strong></li><li><strong>Thin or negative profit margins</strong></li><li><strong>Misalignment of needs</strong>, which may lead to customer dissatisfaction — not because your product is bad, but because it simply wasn’t meant for them</li><li>The risk of <strong>negative word-of-mouth</strong>, which can harm your brand despite the quality of your product</li></ul><p>Examples of this kind of <em>non-marketing</em> include:</p><ul><li>Spending heavily across <em>all advertising channels</em> without segmenting the target audience</li><li>Offering <em>steep discounts</em> just to attract “any” customer, regardless of fit</li><li>Being active on social media platforms that <strong>don’t align with your audience</strong></li><li>Trying to create “brand distinction” through sheer visibility rather than strategic alignment</li></ul><p>These mistakes are unfortunately <strong>common in startup ecosystem</strong> today, where growth at any cost often overshadows the strategic discipline of <em>true</em> marketing.</p><blockquote><strong>In the Case of C<em>artonino</em>…</strong></blockquote><blockquote>Despite having a solid product and a rich library of kids’ movies, your board and internal team are dissatisfied with revenue. After a brainstorming session, you all agree on a bold campaign:<br> → A <strong>20K lottery prize</strong> awarded monthly to one lucky subscriber.</blockquote><blockquote>Now, by simply purchasing a <strong>30$ subscription</strong>, users not only gain access to your extensive children’s content but also get a chance to win a big cash prize. It seems like the <em>perfect</em> campaign to promote a truly valuable product.</blockquote><blockquote>You launch it.</blockquote><blockquote>And in just one month, your sales multiply by <strong>15x</strong>.</blockquote><blockquote>No one doubts that the campaign should continue in the following months — after all, people will eventually “realize the value” of your product, right?</blockquote><blockquote>But then you notice something: The <strong>new users</strong> are <strong>far less engaged</strong> than your original subscribers</blockquote><blockquote>The <strong>churn rate</strong> skyrockets</blockquote><blockquote>You’ve fallen into a trap: <strong>Unmarketing</strong>.</blockquote><blockquote>Many of the people who purchased a subscription <strong>don’t even have children</strong>. Most of them <strong>never watched a single film</strong> on Cartonino. They’re prize hunters — gamblers, sweepstakes chasers who saw your product not as a valuable kids’ VOD service, but as <strong>a lottery ticket</strong>.</blockquote><blockquote>You’ve <strong>sold the product to the wrong customer (uncustomer)</strong>.</blockquote><blockquote>This is <em>Unmarketing</em>.</blockquote><h4><strong>2 — Selling an Unproduct to a Customer</strong></h4><p>Excessive focus on <strong>scalable growth</strong> during the early stages of a startup often leads to offering a product that <strong>hasn’t yet aligned</strong> with real customer needs. What’s advertised as the value proposition doesn’t match what the actual product delivers — and this mismatch creates <strong>disappointment</strong>, eroding any chance of rebuilding trust in the future.</p><p>You’ve likely seen <strong>many examples of this kind of unmarketing</strong> in the rise and fall of companies.</p><h3>What causes it?</h3><ul><li><strong>Growth outpacing product development</strong></li><li><strong>Competitive strategies</strong> overtaking <strong>customer development strategies</strong></li><li><strong>Poor understanding</strong> of the actual customer’s needs</li><li>Product design based on <strong>founder’s instinct</strong> or <strong>blind imitation</strong> of other successful products</li></ul><p>Some startups even raise <strong>large rounds of funding</strong> before achieving <em>product-problem fit</em> — only to <strong>burn through that capital</strong> on unmarketing: <em>pushing an immature, misaligned product into the market too soon.</em></p><p>This is not just bad marketing. It’s <strong>anti-marketing</strong> — because it damages future opportunity as well.</p><blockquote><strong>In the case of Cartonino…</strong><br>Not long after your product’s official launch, your infrastructure hasn’t yet been stress-tested by heavy simultaneous usage. Technically, you’re still in a delicate phase.</blockquote><blockquote>Your marketing team, fresh from a growth hacking event, proposes implementing a growth hack tactic — and you agree.</blockquote><blockquote>Within one week, <strong>simultaneous user activity surges by 750%</strong>, setting a new record. But just <strong>30 minutes later</strong>, your servers <strong>collapse</strong> under the unexpected load.</blockquote><blockquote>Streaming becomes choppy and low-quality, and <strong>everything grinds to a halt</strong>. It takes your tech team <strong>a full week</strong> to resolve the issue — but by then, <strong>many of the new users have already lost trust</strong> in your service during their first critical experience.</blockquote><blockquote>This is a classic example of <strong>unmarketing</strong>: You sold an <strong>unproduct</strong> to <strong>customers</strong>.</blockquote><blockquote>It’s a costly mistake — not just technically, but reputationally. <strong>Regaining trust</strong> from users who were disappointed during their first impression is <strong>incredibly difficult</strong>, and in some cases, impossible.</blockquote><h4><strong>3 — Selling an Unproduct to an Unustomer</strong></h4><p>This is <strong>the fastest route to failure</strong> for any marketing team.</p><p>Marketing teams that are <strong>isolated from the product design process</strong> and are evaluated based on <strong>flawed KPIs</strong> set by management often fall into this trap of <strong>unmarketing</strong>. In this case, the marketing team suffers from a <strong>strategic misunderstanding</strong> — one rooted in a poor grasp of business goals, customer insight, and the product development process.</p><p>To prevent this, a <strong>lean, tightly integrated feedback loop</strong> must exist between the marketing and product teams. Without it, both functions operate on faulty assumptions.</p><p>Furthermore, these teams often lack a <strong>clear market targeting strategy</strong>. Their efforts to attract users are misaligned with actual product-market fit. They assume <strong>everyone is a customer</strong>, and worse — that <strong>every type of customer can be found through every marketing channel</strong>.</p><p>This misunderstanding leads to wasted budgets, confused messaging, and ultimately, a damaged brand with no real traction.</p><blockquote><strong>At Cartonino…</strong><br> Samin’s mother — one of the marketing team members — runs a preschool. She has been using Cartonino during movie-watching hours at her center and has given Samin very positive feedback. Inspired by this, Samin proposes expanding Cartonino’s reach by partnering with preschools across the country.</blockquote><blockquote>However, all your product and market development strategies are built around a <strong>B2C business model</strong>. With this suggestion, the marketing team unintentionally shifts toward <strong>B2B market expansion techniques</strong>. Meanwhile, your product still lacks complete categorization, and supporting <strong>big-screen video streaming</strong> is not yet a development priority.</blockquote><blockquote>Samin’s proposal is a case of <strong>unmarketing</strong>, and by accepting it, you’ve made a <strong>strategy-execution mismatch</strong>. Given your business model, your team is not equipped to efficiently pursue B2B market development. A significant portion of your marketing budget is now being spent on trial-and-error efforts and building a sales unit for preschools.</blockquote><blockquote>These types of operations come with high fixed and upfront costs, and they impose a substantial <strong>opportunity cost</strong> by diverting focus away from the core B2C efforts. Meanwhile, the current version of Cartonino is not a good fit for the preschool use case — it lacks the stickiness and effectiveness required to retain these institutional clients. Over time, they grow indifferent to your service.</blockquote><h4><strong>4 — Unselling the Product to the Customer</strong></h4><p>This type of unmarketing refers to missed sales opportunities — instances where the right product isn’t sold to the right customer for various reasons. Common causes include absence from effective marketing channels where target customers are actually present, lack of clarity around who the ideal customer is, and failure to segment customers based on profitability.</p><p>Customer segmentation based on profitability and behavior analysis can clearly reveal how much the marketing team is falling into this trap and what adjustments are needed. On the other hand, a lack of boldness in the marketing team to innovate or “<strong>think outside the box</strong>” when selecting diverse marketing channels can eventually result in the right product not being promoted to the right customers.</p><blockquote><strong>In the case of Cartonino…</strong></blockquote><blockquote>To run display advertising for Cartonino, the marketing team experiments with several different channels. One of these is the <em>“NiniSite — Regional Section”</em>, a banner ad network that targets users living outside of capital and charges on a CPM basis. After evaluating performance, the team discovers that the customer acquisition cost (CAC) in this channel is four times higher than in others. As a result, they decide to stop running ads in this channel. This decision also happens to align with the company’s market expansion strategy for the first year, which focuses on Capital.</blockquote><blockquote>A few months later, an eRFM-based customer profitability analysis reveals that a small segment of users has a much higher customer lifetime value (CLV) — eight times the average. Further research, including surveys and interviews, shows that these high-value users mostly live outside Capital. Due to limited access to physical cartoon DVDs, they have a much stronger preference for online streaming services like Cartonino.</blockquote><blockquote>By not maintaining a presence in geo-targeted advertising channels, the team has fallen into a unmarketing trap. Making decisions based solely on CAC, without considering CLV, has resulted in missing out on more valuable customers.</blockquote><p>As you can see, only in the fourth scenario are marketing and sales aligned. In the first three scenarios — despite successful sales — what actually occurred was <em>unmarketing</em>. This is the most important insight every marketer must understand and apply within their own business context. What percentage of your marketing activities is actually <em>marketing</em>, and what percentage is <em>unmarketing</em>?</p><blockquote>Deciding where to invest your time, money, and energy — on which innovations to pursue — is a fundamental part of the marketing process.<br> ~ William A. Cohen, <em>Peter Drucker on Marketing</em></blockquote><h3><strong>4. When Organizational Culture Eats Strategy for Breakfast</strong></h3><p>Following these principles requires an organizational culture that goes beyond profit-seeking. According to Peter Drucker, the companies that have <strong>managed to survive</strong> in the long term are those where a culture of putting the customer before the organization has been deeply embedded.</p><p>This should not be confused with customer service or customer-centric slogans. Prioritizing the customer in organizational culture means placing the customer’s interests ahead of sales and revenue. It means that when you’re faced with a choice between what’s best for the customer and what’s best for the company, you choose the customer.</p><p>Managers should not evaluate the performance of the marketing team solely based on metrics like total revenue or number of sales. If they do, the marketing team effectively becomes a sales team — focused on generating short-term income rather than building long-term customer value (and sustainable profitability).</p><p>Let’s go back to the story at the beginning: that outdoor gear salesperson, when realizing that what benefits <em>you</em> — both financially and practically — is buying a cheaper tent, has a choice. Even though selling you a more expensive tent (which you’re actually leaning toward due to lack of knowledge) would benefit him more, he chooses the first option. He sells the cheaper tent, creating a more satisfied customer in the process. He is selling the right product to the right customer (<em>marketing</em>), and avoiding selling the wrong product to the wrong customer (<em>unmarketing</em>).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PERHKzYmLZMjIsmYDRWVkQ.png" /></figure><h4><strong>So, What Should We Do? Sell or Not Sell?</strong></h4><p><strong>First</strong>, take time to really reflect on this article. Share it with your teammates and have an open discussion about it. If you agree with the ideas presented here, try identifying specific examples from your team’s past marketing activities using the <em>Marketing–Unmarketing</em> graph. Talk through what these concepts actually mean in the context of your business.</p><p><strong>Second</strong>, think deeply about the concept of <em>creating customers</em>. Bring up examples where you may have prioritized customer benefit over the organization’s — or vice versa. Use these discussions to define clear principles and guidelines tailored to your business model, and make sure these are shared across the team.</p><p><strong>Third</strong>, when evaluating your marketing team’s performance, avoid relying on misleading <em>vanity metrics</em>. Instead, focus on <em>practical metrics</em> — those that directly measure the effectiveness of your marketing efforts and the product itself. Metrics such as churn rate, retention rate, and customer return rate are far more useful for understanding real impact.</p><h4><strong>If you’ve stayed with me this far…</strong></h4><p>The core message of this article could have easily been condensed into a short tweet about that camping tent purchase. But I chose to share it in the most detailed way I could manage, given the time and energy I had. My goal was to ensure we don’t gloss over this fundamental principle of marketing. By introducing concepts like U<em>nmarketing</em>, Un<em>product</em>, and <em>Uncustomer</em>, I hoped to encourage deeper and more effective conversations in our professional circles.</p><p>If you learned something new from reading this:</p><ul><li><strong>First</strong>, please consider sharing it with others. That way, your friends and colleagues might avoid making similar unmarketing mistakes.</li><li><strong>Second</strong>, share examples of <em>marketing</em> or un<em>marketing</em> you’ve seen — in your own experience or in other businesses — in the comments section. Your stories will help others better understand these concepts in a real-world context.</li></ul><p>And if you have any <strong>criticisms or suggestions</strong> for improving this piece, don’t hesitate to let me know. Your feedback can help ensure that others don’t misunderstand these ideas.</p><p>A significant part of my understanding of Peter Drucker’s theories on marketing comes from reading this book. For anyone currently immersed in strategy, marketing, and growth, I highly recommend reading: <strong>“Drucker on Marketing: Lessons from the World’s Most Influential Business Thinker”</strong> by <em>William A. Cohen</em></p><blockquote><em>Additional tools and resource ➡️ </em><strong><em>Visit </em></strong><a href="http://www.startupstash.com/"><strong><em>StartupStash</em></strong></a><strong><em><br></em></strong><em>Zendesk is giving $75,000 in credits and perks for startups! ➡️ </em><a href="https://tinyurl.com/4ta2c8j6"><strong><em>Apply Now!</em></strong></a></blockquote><p><em>This article was originally written in Persian, published in </em><a href="https://vrgl.ir/1ViP8"><em>virgool</em></a><em>. I translated it into English with the assistance of ChatGPT.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cd5ea74f6f47" width="1" height="1" alt=""><hr><p><a href="https://blog.startupstash.com/unmarketing-unproduct-uncustomer-cd5ea74f6f47">Unmarketing, Unproduct, Uncustomer</a> was originally published in <a href="https://blog.startupstash.com/">Startup Stash</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>