<?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 Ilona Hetsevich on Medium]]></title>
        <description><![CDATA[Stories by Ilona Hetsevich on Medium]]></description>
        <link>https://medium.com/@ihetsevi?source=rss-f44c0bb08544------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*GzsZBZVN_fq0r_mNuu87vw.jpeg</url>
            <title>Stories by Ilona Hetsevich on Medium</title>
            <link>https://medium.com/@ihetsevi?source=rss-f44c0bb08544------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 22:06:12 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@ihetsevi/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[Stop Making Reports No One Reads: How to Build Data Decks That Drive Action]]></title>
            <link>https://medium.com/data-and-beyond/stop-making-reports-no-one-reads-how-to-build-data-decks-that-drive-action-ab9bec9ba2e9?source=rss-f44c0bb08544------2</link>
            <guid isPermaLink="false">https://medium.com/p/ab9bec9ba2e9</guid>
            <category><![CDATA[storytelling]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[data-visualization]]></category>
            <category><![CDATA[data-storytelling]]></category>
            <dc:creator><![CDATA[Ilona Hetsevich]]></dc:creator>
            <pubDate>Tue, 17 Mar 2026 14:40:04 GMT</pubDate>
            <atom:updated>2026-03-17T14:40:04.887Z</atom:updated>
            <content:encoded><![CDATA[<h4>Weeks of hard work shouldn’t end in a forgotten PowerPoint.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5X3AFq48IPj4BSjOUAQCbg.png" /></figure><p>If you are anything like me, you have probably lost count of the reports, readouts, deep-dives, scorecards, and QBRs you’ve created during your data analyst career. And how many of those were forgotten and never acted on?</p><p>I bet there are many.</p><p>A deck is often the final output of the analysis and the result of weeks of hard work. Yet, we, as data analysts, don’t spend much time learning how to do it effectively, and as a result, our recommendations never make it off the slide.</p><blockquote>“Why bother if we can use AI to create a deck for us?” you might ask.</blockquote><p>It’s a fair question.</p><p>While AI can definitely be a powerful drafting partner, you are the one with a human perspective. Moreover, the ability to present insights clearly and create decks that advance your ideas and recommendations can set you apart professionally and help you avoid being replaced by AI.</p><p>I created this guide based on 13 years of experience working with data and presenting the insights, as well as multiple books I’ve read and courses I’ve taken throughout my career.</p><blockquote><strong>Disclaimer:</strong> There are multiple deck formats, and the approach will differ depending on whether you are working on a deck to present the results of deep-dive analysis, a templatized QBR (Quarterly Business Review), an executive scorecard, or a regular performance report. For this article, I focus particularly on the first: creating a deck to present deep-dive analysis insights.</blockquote><p>So, what are we waiting for — let’s dig in:</p><h3><strong>Where to start?</strong></h3><p>You’re done with the analysis. Weeks of hard work are behind you. And now you are staring at an empty PowerPoint deck, not knowing how to fit weeks’ worth of data analysis into just a few slides.</p><p>It feels impossible, but it’s not — here’s how it’s done:</p><h3><strong>Structure of the Deck</strong></h3><p>Through a lot of trial and error over the years, I’ve come up with a template I usually follow — and it goes like this:</p><ul><li>The deck title, author, and date</li><li>Purpose of the analysis</li><li>Data sources, timeline, scope, methodology, data limitations, definitions <em>(you can either put them on one slide or have multiple)</em></li><li>Executive Summary</li><li>The slides to present the insights</li><li>Recommendations</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xanKY3aBxBOIuFQMg1-PaA.png" /></figure><p>I always tailor my content to my audience — i.e., what they already know about the topic, the available time, and whether they’ll be reading or watching — but the structure usually stays unchanged. What also stays the same is that I strive to make the deck self-explanatory, even if I am not in the room to narrate and present.</p><blockquote><strong>Anyone who gets access to the deck afterward will need to be able to get to the main point without needing to watch a recording.</strong></blockquote><p><strong>The first 3 slides</strong> (deck title, goal of the analysis, data sources, timeline, scope, methodology, and definitions) are there to set the context and answer all the initial data foundation questions. If you have been in a data analyst role for a while, you know how important the context is and how drastically the insights can change depending on the data sources you use or the timeframes you include.</p><blockquote>And there will always be someone questioning your data, especially if your conclusions “challenge the status quo” or question someone else’s work effectiveness.</blockquote><p>Thus, I never omit these slides, even though I rarely present them.</p><p><strong>The Executive Summary</strong> is a must-have slide for everyone who is “too busy” to go through the entire deck. Along with your Recommendations, these are the two most vital slides in your deck.</p><p>Even if you have only two minutes of stakeholder attention, these two slides will get your message across and ensure your work makes an impact. If you must cut everything else, these 2 slides must stay.</p><p>I recommend spending as much time on them as on all the other slides covering the insights. I place the Executive Summary slide at the beginning of the deck, but I always create it last. This way, I am sure nothing is missed.</p><h3><strong>The Slides to Present the Insights</strong></h3><p>Data storytelling books usually recommend outlining the story structure before creating slides. While I do understand the point (you don’t want to waste your time creating slides you might not need), I approach this a little differently. I’ll normally put together a rough, messy draft first without worrying too much about the narrative or overwhelming the audience with too many metrics and then build the story and decide what stays and what goes to the appendix.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/802/1*sST63UnEdrHCUVyxfnH9qg.png" /></figure><p>When organizing slides, I usually follow <strong>a vertical flow:</strong> all the content on the slide should relate to its title.</p><p>Slide structure varies, but usually, there is</p><ul><li>Headline</li><li>Sub-headline</li><li>Visual(s) or Table</li></ul><p>Let’s start with the slide headline:<strong> a<em>n audience</em> should be able to get the message you are trying to convey just from reading the headline.</strong></p><blockquote><em>If you read only the slide titles, would the story still make sense?</em></blockquote><p>I avoid generic headlines like “Marketing Sourced Win Rate Trendline.” Instead, I use the headlines that explain exactly what happened: “<em>Q2 Marketing Sourced Win Rate dropped to 12.6% with big deals pushed to next quarter.”</em></p><p>Usually, I also add a sub-headline — 1–2 lines of text that provide extra details or supplementary context not in the data. For example,<em> “20 large deals (about $4.5M total) slipped to next quarter. Buyers requested a short pilot to test integrations.”</em></p><p>Last but not least, I never include multiple messages on the same slide. <em>(Do not confuse it with multiple charts per slide. It is totally fine to have more than 1 visualization on the slide if visuals are connected and help you communicate the main message).</em></p><p>Each slide needs to transition smoothly into the next one (<strong>a horizontal flow</strong>). Like classical storytelling in books and movies, when creating a deck, I follow a narrative arc:</p><ol><li>First, reveal the facts;</li><li>In the middle, “investigate” the barriers;</li><li>At the end, propose a way forward.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Tsjwqgw3IYgUqbHUvatCmg.png" /></figure><p>Familiarity with the approach makes it intuitive for the audience to follow.</p><h3><strong>And Now the Visuals</strong></h3><p>Always keep in mind that the chart that helped you discover the insight and the chart you include in the deck are two different charts. For example, the discovery chart might show 12 channels plotted in multiple colors, with overlapping lines. It helps you identify trends and spot the main drivers. But if left unchanged on the slide, it will overwhelm the audience. The presentation chart should focus on the main drivers only, with the rest greyed out or removed.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4qcnUCeZSlLzrBBwyKUpJw.jpeg" /></figure><h4><strong>Use Color Effectively</strong></h4><ul><li>To effectively guide your audience’s focus and highlight the key message, use color, contrast, and line thickness.</li><li>Another point about color is consistency. If you’ve used green color for the Paid Search channel on one slide, keep using the same green color in all the charts with Paid Search.</li><li>A colorblind-friendly palette is usually a safe choice.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/704/1*sBxRBF_82d7pPsbH0OTD1Q.png" /></figure><h4><strong>Focus on Readability</strong></h4><ul><li>Increase the gap between bars to improve readability.</li><li>Remove clutter and distractions from your visualizations (gridlines, redundant axes, unnecessary markers, data labels).</li><li>Replace crowded small intervals with broader timeframes to highlight big-picture trends.</li><li>Improve the x-axis readability by separating the year and quarter into distinct levels.</li><li>Do not rotate the labels.</li><li>Use visual hierarchy, sort in descending order.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/695/1*p-yZngZqTnSxKLN50-kQrQ.png" /></figure><h4><strong>Aim to Decrease Cognitive Load</strong></h4><ol><li>Do math to reduce cognitive load for the audience.</li><li>Replace busy dual-axis chart with 2 separate ones as the different units of measures and scales can make it tricky to identify meaningful trends.</li><li>Label the lines directly in the chart, so the audience won’t need to check the legend back and forth.</li><li>Format large numbers for readability (for example, 100K instead of 100,000).</li><li>When using a table, consider grouping related rows and columns, or applying background colors for a quicker review.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/716/1*vTzFW0v1YJ4QSiQJombyVg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/702/1*V7N7wZelXAMiYKyEZ1tEdg.png" /></figure><h3><strong>Recommendations</strong></h3><p>When I build a recommendation slide, I usually follow the <strong>“What — So what — Now what”</strong> structure.</p><ul><li>“<strong>What” </strong>is the insight I’ve discovered.</li><li>“<strong>So what”</strong> is the reason why it is important.</li><li><strong>“Now what”</strong> is what we should do about it. Actionable recommendation (ideally time-bound, connected to revenue, with team/person responsible indicated).</li></ul><h4>Example:</h4><p><strong>What:</strong> Q2 Marketing Sourced Win Rate dropped to 12.6%, with 20 large deals (about $4.5M total) pushed to next quarter after buyers asked for pilot integrations.</p><p><strong>So what:</strong> This dip in win rate and the delay in closing puts our quarterly revenue target at risk and hints at friction late in our sales process.</p><p><strong>Now what:</strong> Collaborate with sales, solutions, and other teams to speed up and simplify pilots as well as create fast-start onboarding materials. The goal is to shorten deal cycles and win back those deals that have slipped into Q3.</p><h3><strong>Appendix</strong></h3><p>Once you deep-dive into the analysis, you discover many findings that are relevant to a story but are not necessarily worth sharing during a presentation, or their impact is minimal. This is what the Appendix is for. And the slides here are not buried; quite the opposite.</p><p>You don’t know how many times the slides in the Appendix saved me when answering the audience’s questions.</p><blockquote>Trust me, there is always someone in the audience who wants to know more, dig deeper, or ask ‘have you checked…’ types of questions.</blockquote><h3><strong>Final Thoughts</strong></h3><p>So, you now know how to put together a well-organized and impactful deck.</p><p><strong>The last thing I want to mention</strong> is that you also need to practice the deck delivery. You can build the most amazing deck ever, but if your speech is monotonous, has too many fillers (<em>um/uh/like</em>), if you are speaking too fast or too slow, if you are mumbling or just reading the text from a slide, who can blame people for not listening to you?</p><p>Practice-practice-practice!</p><p>Thank you for reading the article to the end. If you like what you’ve read and would like to stay in touch, feel free to <strong>connect with me on </strong><a href="https://www.linkedin.com/in/ilonahetsevich/"><strong>LinkedIn</strong></a><strong>.</strong></p><p><em>All images were created by the author.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ab9bec9ba2e9" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-and-beyond/stop-making-reports-no-one-reads-how-to-build-data-decks-that-drive-action-ab9bec9ba2e9">Stop Making Reports No One Reads: How to Build Data Decks That Drive Action</a> was originally published in <a href="https://medium.com/data-and-beyond">Data And Beyond</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The New Analytics Trap: When Everyone Can Analyze Data, Fewer People Tell the Truth]]></title>
            <link>https://medium.com/data-and-beyond/the-new-analytics-trap-when-everyone-can-analyze-data-fewer-people-tell-the-truth-6d6e9364b04d?source=rss-f44c0bb08544------2</link>
            <guid isPermaLink="false">https://medium.com/p/6d6e9364b04d</guid>
            <category><![CDATA[analytics]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[marketing-analytics]]></category>
            <category><![CDATA[data]]></category>
            <dc:creator><![CDATA[Ilona Hetsevich]]></dc:creator>
            <pubDate>Mon, 02 Mar 2026 08:27:48 GMT</pubDate>
            <atom:updated>2026-03-02T08:27:48.100Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Why your “data-driven” marketing slides are probably wrong.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qpSBZeeLU7WfrEiHCcdo0Q.jpeg" /><figcaption><em>The image was created by the author using AI.</em></figcaption></figure><p>It’s QBR time. A marketer presents the performance results. Somewhere in the meeting, there is a slide with a big, confident headline: <strong>“Event Attendance increases pipeline value by 35%</strong>.” In the bottom left corner, a short sentence explains the methodology: <em>“Compared the average pipeline value of accounts who attended vs accounts who did not.”</em> The slide is followed by a heated discussion of how we can double down on the events budget and replicate the results in EMEA.</p><p>And just like that, a decision was made based on data easily accessible to everyone, yet it isn’t entirely true. You — a data analyst — are lucky if you are invited to the call in the first place and can speak up. More than ever, these decisions are made behind closed doors — or rather, inside locked virtual meeting rooms — without ever consulting you.</p><p>Today, everyone can generate an analysis by just uploading a CSV export to an AI model. We shifted from waiting weeks for an analyst to crunch the numbers and send out the report to asking an LLM to do it in a matter of minutes. To marketers it is a huge win, for analysts — a headache.</p><p>When everyone can analyze data, we get insights faster, but we also make mistakes more quickly. <strong>That is a hard truth about making data widely accessible.</strong></p><p>So how do we fix it? Let’s dig in.</p><h4><strong>Why Do We Run Into This Problem at All?</strong></h4><p>AI has made analytics feel like a conversation:</p><ul><li>“Calculate the pipeline for accounts that attended an event.”</li><li>“Compare it to accounts that didn’t.”</li><li>“Summarize the insight for leadership.”</li></ul><p>The numbers in the output are correct; nobody argues. The LLM did its job.</p><p>But the real difficulty in analytics was never about getting accurate numbers. The hard part is deciding whether the logic matches a reality:</p><ul><li><strong>There is a selection bias:</strong> event attendance isn’t random; it is heavily shaped by targeting, ICP, ABM tiers, and even handpicked by sales. The accounts that attend are frequently higher-fit and higher-intent before they ever set foot in the room. Of course, they would generate a higher pipeline.</li><li><strong>There are hidden variables</strong> that affect both the likelihood of attending the event and the pipeline outcome: prior engagement, funnel stage, open opportunities already in motion, sales and ABM coverage, marketing spend, existing relationship, account size… the list is long.</li><li>In many B2B companies, <strong>the causality runs the other way:</strong> when accounts are in the buying process (have an active pipeline), sales reps push harder to get them invited to events and dinners exactly because it is a strategic deal. They don’t want to lose it.</li><li>When running fast analysis, <strong>non-analysts tend to make timing mistakes</strong>: reporting Q2 pipeline for accounts who attended an event in Q2, when in reality the pipeline created in Q2 was attributed to marketing activities months/quarters ago and has nothing to do with the event in Q2.</li></ul><p>You see the pattern here?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Q29QBmxsiJVwG6lEaZvwqA.png" /><figcaption><em>The image was created by the author using AI.</em></figcaption></figure><h4><strong>The Hidden Dangers of Self-Serve Analytics</strong></h4><p>Marketers have domain expertise; they know the context behind marketing campaigns and why the numbers have changed (like an increased promo budget or a change in a demand generation strategy). But they are not analysts. They are not supposed to know all the pitfalls of running a data analysis, definitions, grain, cohorts, time windows, exclusions, business context, and all the messy edge cases. And when they rely on AI to deliver insights, they risk building a marketing strategy based on something that never existed.</p><blockquote>The events slide is a perfect case study because it shows the new analytics trap: the LLM’s output seems confident, the numbers align, and the insight must be valid. But it is not.</blockquote><p>The same pattern shows up everywhere in B2B marketing analytics:</p><ul><li>Claiming ABM increased pipeline when comparing ABM-targeted Tier 1 accounts to non-targeted accounts.</li><li>Interpreting a spike in paid leads as the driver of a pipeline spike in the same quarter.</li><li>Stating “more spend brings more leads and thus more revenue” without separating volume from efficiency and incrementality.</li><li>Assuming conversion lift is caused by the one variable measured instead of a mix of things happening at the same time.</li><li>Etc.</li></ul><p>You can input accurate data and still get a confident AI-generated insight that is wrong because the framing ignores targeting, omitted variables, selection effects, and timing.</p><p><strong>The mistakes made by non-analysts are not random</strong>. Usually, I see the same mistakes again and again:</p><ul><li>Declaring that “X” caused “Y” from a trend.</li><li>Slicing and dicing the data multiple ways until something looks good enough to share with leadership.</li><li>Making decisions using statistically insignificant data.</li><li>Picking an A/B test winner from a small sample size.</li><li>Not speaking the same language — marketing says Churn is 2% and Product says it’s 5% — when in reality they use a different definition of Churn in the first place.</li><li>Ignoring seasonality, comparing data QoQ instead of YoY.</li><li>Analyzing data in a vacuum, disregarding direct and indirect competitors’ activity, product releases, economic changes, etc.</li><li>Not being aware of changes in the data underlying data sources, in the attribution modeling, the addition of new data sources, and tracking issues.</li><li>And more.</li></ul><h4><strong>What Happens When Insights Go Wrong</strong></h4><p>When insights go wrong, teams end up making the wrong calls, budgets get misallocated, and revenue is left on the table. But the damage doesn’t stop there — <strong>there are ripple effects beyond just the numbers</strong>:</p><ul><li><strong>Leadership stops trusting the data </strong>and starts questioning every insight, asking to validate it with the data team. As a result, “speed to insight” loses its meaning, because the analyst needs to spend time validating the results, even when it wasn’t necessary in the first place.</li><li>When there are multiple people running analyses, delivering multiple “truths”, the story by the most persuasive narrator, with the strongest voice and <strong>most plausible interpretation, wins</strong>.</li><li><strong>The relationship between marketers and analysts falls apart</strong>. Rather than working together, they see each other as rivals. Marketers feel they are being blocked and not trusted. While analysts feel their analytics experience is used to “babysit” and debug someone else’s work that is clearly generated by AI, rather than producing real value for the business.</li><li><strong>“Shadow analytics” grows</strong>. People start using private spreadsheets and unapproved AI tools. Cross-departmental communication breaks down, nobody trusts the data, and everyone leaves in their own reality, with their own metrics and insights.</li></ul><h4><strong>What Can We Do to Fix It?</strong></h4><p>Of course, we cannot “lock down” data. We need to build a new way to work together.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2FO_OplnnOoG_zGDM6QcrQ.png" /><figcaption><em>The image was created by the author using AI.</em></figcaption></figure><p>Here are 3 options to handle it:</p><ol><li><strong>Separate using AI for exploration from reporting presented to leadership.</strong></li></ol><p>Give people freedom to explore data with AI, find trends, brainstorm together, and make low-stakes internal decisions. But require validation by the data team for anything presented to leadership, especially when it involves high-stakes or budget decisions. Their review helps catch mistakes and makes sure the numbers can be trusted.</p><p><strong>2. Build guardrails to make self-service analytics safe.</strong></p><p>What you can do, if not already done, is to build and maintain a clear data dictionary so everyone uses the same definitions for metrics, and the next time they are in the same room, they talk about the same thing when referring to “Churn”, “Conversion”, or “Engagement”.</p><p><strong>3. Ensure marketers and analysts partner together.</strong></p><p>Analysts should attend marketing calls regularly to stay in the know about what is happening in marketing — latest strategy changes, new campaign launches, budget adjustments — so they can explain the changes in numbers. At the same time, marketers need to understand the data before running any analysis. If a marketer cannot explain how the metric is defined, they are not ready to share the insights. What works is having regular office hours for marketers to ask questions and validate their hypotheses.</p><h4><strong>Final Thoughts</strong></h4><p>Data belongs to everyone now.</p><p><strong>For Marketers</strong>, this is a call to improve your skills. Prompt engineering alone isn’t enough. You need data literacy; you need to learn how to spot when something is off or too good to be true.</p><p><strong>For Analysts</strong>, this is a call to let go. You can’t check every piece of data that goes out. Your value now comes from designing a framework that makes self-service analytics safe.</p><p>In the age of AI, the data itself is not particularly valuable; what matters is that you can trust the data and the insights derived from it.</p><p>Thank you a lot for reading the article to the end. If you like what you’ve read and would like to stay in touch, feel free to connect with me on <a href="https://www.linkedin.com/in/ilonahetsevich/">LinkedIn</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6d6e9364b04d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-and-beyond/the-new-analytics-trap-when-everyone-can-analyze-data-fewer-people-tell-the-truth-6d6e9364b04d">The New Analytics Trap: When Everyone Can Analyze Data, Fewer People Tell the Truth</a> was originally published in <a href="https://medium.com/data-and-beyond">Data And Beyond</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Behind the Scenes of a Successful Data Analytics Project]]></title>
            <link>https://medium.com/data-science/behind-the-scenes-of-a-successful-data-analytics-project-8d45fb33e9c5?source=rss-f44c0bb08544------2</link>
            <guid isPermaLink="false">https://medium.com/p/8d45fb33e9c5</guid>
            <category><![CDATA[data-analyst]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[data-analytics]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[project-management]]></category>
            <dc:creator><![CDATA[Ilona Hetsevich]]></dc:creator>
            <pubDate>Thu, 23 Jan 2025 11:32:03 GMT</pubDate>
            <atom:updated>2025-01-23T11:32:03.099Z</atom:updated>
            <content:encoded><![CDATA[<h4><em>Learn the steps to approach any data analytics project like a pro.</em></h4><p>Having worked as a data analyst for a while and tackled numerous projects, I can say that even though each project is unique, there is always a proven way to approach it.</p><p>Today, I’ll share with you the steps I usually take when working on a data project so you can follow them too.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fhfqsp4JyfCTIlwQgPTuqA.png" /></figure><p><strong>Step 1: Define the Problem and Objectives</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/519/1*2_X0Ko40dHkkcmAh5se6Ng.png" /></figure><p>You cannot solve a problem or answer a business question if you do not understand what it is and how it fits into the bigger picture.</p><p>No matter how big or complex the task is, you must always understand what your business stakeholders are trying to achieve before diving into data. This is the part where you ask many questions, and before you get at least some answers, you are not diving into any data.</p><p>I learned this the hard way early in my career. Back then, when a vague request like <em>“We saw visitors drop this month. Can you check why?”</em> came, I would immediately jump into work. But every single time, I wasted hours trying to understand the real problem because I didn’t ask the right questions upfront.</p><p><strong>I didn’t ask for context:</strong></p><ul><li>Why the team needed the traffic to be high?</li><li>What was the chosen strategy (brand awareness vs demand generation)?</li><li>What were the chosen tactics (paid search vs programmatic)?</li><li>What were the investments?</li></ul><p><strong>I didn’t ask stakeholders what they would do after receiving the data.</strong></p><ul><li>Did they want to increase signups and sales?</li><li>Were they aware that website visits may look impressive but not necessarily correlate with business outcomes and that focusing on metrics such as conversion rate would have a much better effect?</li></ul><p>This initial first step is important because it affects everything else: from the <strong>data sources</strong> you will use to retrieve the data to the <strong>metrics</strong> you will analyze, the <strong>format</strong> you will use to present the insights, and the <strong>timeline</strong> you need to be ready for.</p><p><strong>So don’t ever skip it</strong> or just partially understand hoping you will figure it out along the way.</p><p><strong>Step 2: Set Expectations</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/523/1*SRGrIm2kPZ82l4bSyv4TQA.png" /></figure><p>Once you’ve defined the problem, it’s time to set expectations.</p><p>Stakeholders don’t always realize how much time and effort goes into collecting and analyzing data. You are among the few people in the organization who can find the answers, so you receive many requests. That is why you need to prioritize and set expectations.</p><p>Understanding the problem, its complexity, how it aligns with the organization’s goals from Step 1 helps prioritize and communicate to stakeholders when the task can be done or why you will not be prioritizing it right now. You want to focus on the most impactful work.</p><p>A colleague of mine took a smart approach. They required stakeholders to fill out a questionnaire when submitting a task. This questionnaire included various questions about the problem description, timeline, etc., and it also asked, “<em>What will you do with the insights?</em>”. This approach not only gathered all the necessary information upfront, eliminating the need for back-and-forth communication, but it also made stakeholders think twice before submitting another <em>“Can you quickly look at…?”</em> request. Genius, right?</p><p><strong>Step 3: Prepare the Data</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/515/1*LXFhAGK3ROAafsPHaH6zRg.png" /></figure><p>Now that you’ve defined the problem and set expectations, it’s time to prepare the data.</p><p>This is the step where you ask youself:</p><ul><li>Do I have all the data available, or do I need to collect it first?</li><li>Do I have all the domain knowledge needed, or do I need to do the research?</li><li>Do I have documentation available for the associated datasets? <em>(If there is no documentation, you may need to contact the data owners for clarification.)</em></li></ul><p>Another critical question to answer at this step is <strong><em>“</em>What metrics should I measure?”</strong></p><p>I always align my metrics with the business objectives. For instance, if the goal is to increase brand awareness, I prioritize metrics like impressions, branded search volume, direct traffic, and reach. If the objective is to drive sales, I focus on conversion rates, average order value, and customer acquisition cost. I also explore secondary metrics (demographics, device usage, customer behavior) to ensure my analysis is comprehensive and paints a complete picture.</p><p><strong>Step 4: Explore the Data</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/524/1*j4mUvp0_6MfejQkDHJCxkw.png" /></figure><p>Now comes the fun part — Exploratory Data Analysis (EDA). I love this part because it is where all the magic happens. Like a detective, you review the evidence, investigate the case, formulate hypotheses, and look for hidden patterns.</p><p>As you explore the data, you:</p><ul><li><strong>Ask better questions.</strong> As you become more familiar with data, you can approach data owners with concrete questions, making you look competent, knowledgeable, and confident in the eyes of your colleagues.</li><li><strong>Innovate with feature engineering. </strong>You understand whether or not you need to create new features from existing ones. This helps to better capture the underlying patterns in the data that would otherwise go unnoticed.</li><li><strong>Assess data quality. </strong>You check the number of rows of data and whether there are any anomalies, such as outliers, missing, or duplicate data.</li></ul><p>If the exploration step shows data needs to be cleaned (and believe me, it is more than the case that it is not), you proceed with data cleanup.</p><p><strong>Step 5: Clean the Data</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/518/1*lEGXs8oA3H7x25k30mVLGQ.png" /></figure><p>No matter how polished a dataset looks at first glance, never assume it’s clean. Data quality issues are more common than not.</p><p>The most common data quality problems you need to fix are:</p><p><em>1. </em><strong>Missing values:</strong></p><p>The way you will handle missing data differs from case to case.</p><ul><li>If it is due to errors in data entry, collaborate with the relevant teams to correct it.</li><li>If the original data cannot be recovered, you need to either remove missing values or impute them using industry benchmarks, calculating the mean or median, or applying machine learning methods.</li><li>If missing values represent a small portion of the dataset and won’t significantly impact the analysis, it is usually OK to remove them.</li></ul><p><em>2. </em><strong>Inconsistent data: </strong>Check data for inconsistent data formats and standardize them.</p><p><em>3. </em><strong>Duplicate records: </strong>Identify and remove duplicate records to avoid skewing results.</p><p><em>4. </em><strong>Outliers or errors in data: </strong>Check for outliers or errors in the data. Based on its context, decide whether to remove, fix, or keep it.</p><p>Once your data is cleaned, it is time to proceed to the analysis phase.</p><p><strong>Step 6: Analyze the Data</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/518/1*P_5UTUDoXIZ2FymbFmEozg.png" /></figure><p>This is where your detective work starts to pay off.</p><p>The key is to start with a very focused and specific question and not to be biased by having a hypothesis in mind. Using data to tell the story you or your colleagues want or expect to hear might be tempting, but you must let the dataset speak for itself.</p><p>I prefer to use the root-cause approach when analyzing data. For example, to answer the question, <em>“Why do we see a drop in signups?”</em> I would follow these 10 steps:</p><ol><li><strong>Trend analysis: </strong>When does the drop happen for the first time? Is it seasonal?</li><li><strong>Traffic and conversion rates:</strong> Are fewer people visiting the site or fewer visitors signing up?</li><li><strong>Offer performance: </strong>Is the decline widespread or isolated to a particular offer?</li><li><strong>Website Performance:</strong> Are there any technical issues or broken links?</li><li><strong>User insights:</strong> Is the pattern specific to a particular segment or all users?</li><li><strong>User journey analysis: </strong>Are there any friction points where potential customers drop off?</li><li><strong>Campaign performance: </strong>Have any recent marketing campaigns or changes in strategy, budget allocation, or execution impacted effectiveness?</li><li><strong>Competitor activity:</strong> Have competitors launched a marketing campaign, new product, or feature? Have they changed their prices? Is there another reason that might be attracting customers away?</li><li><strong>Market trends: </strong>Are there market trends and changes in consumer behavior affecting sales in the industry?</li><li><strong>Customer feedback:</strong> Are customers dissatisfied with the offering? Did their needs change? Do we receive more support tickets?</li></ol><p>Another important point is that the fastest and most accurate answers aren’t usually the same, and a lot depends on<strong> the context. </strong>That is why you need to collaborate with cross-functional teams and develop strong domain and industry knowledge.</p><p><strong>Step 7: Build the Story</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/518/1*Nh-4TZbkBImnrUuqcMo0ug.png" /></figure><p>This step is my second favorite after data exploration because it is when all the data pieces fall into place, revealing a clear story and making perfect sense.</p><p>A common mistake here is including everything you found interesting instead of focusing on what the audience cares about. <br>I get it. After working hard to get insights, it’s tempting to show off all the cool stuff you did. But if you overload your audience with data, you can further confuse them.</p><p>Don’t throw every data point at stakeholders; focus on what matters most to your audience instead. Think about their level of seniority, how familiar they are with the topic, their data literacy level, how much time they have, and whether you’re presenting in person or sending a report via email. This way you don’t waste anyone’s time — yours or theirs.</p><p>Lastly, always include actionable recommendations to stakeholders in your story. Your story should guide stakeholders on the next steps, ensuring that your insights drive meaningful decisions.</p><p>This brings us to the next point — sharing the insights and recommendations.</p><p><strong>Step 8: Share the Insights</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/522/1*AR-rK9g7n4hfO9uVax-ngw.png" /></figure><p>As a data analyst, you have the power to drive change. The secret lies in how you share data and tell the story.</p><p>First, consider the format your audience expects (see Step 1). Are you creating a dashboard, emailing a report, or presenting in person? Data storytelling becomes crucial for live presentations.</p><p>A great data story blends<strong> data, narrative, visuals, and practice:</strong></p><p><strong>Data</strong>: Focus only on insights with real business impact. If you can’t find a compelling reason why your insight will matter to the audience, if it’s unclear what they should do with the insights, or if the business impact is minimal, move it to the appendix.</p><p><strong>Narrative: </strong>Ensure that your story has a clear structure.</p><ul><li>Set the scene: What’s happening now?</li><li>Introduce the problem (to create some tension).</li><li>Reveal the key insights: What did you discover?</li><li>Finish with actionable steps: What should they do next?</li></ul><p>This keeps your audience interested and makes your story memorable.</p><p><strong>Visuals: </strong>The chart that helped you discover an insight isn’t always the best for presenting it. Highlight the key points and avoid clutter. For example, if you analyzed 10 categories but only 2 are critical, focus on those.</p><p><strong>Practice: </strong>Practicing helps you feel more comfortable with the material. It also allows you to focus on important things like eye contact, hand gestures, and pacing. The more you practice, the more confident and credible you will appear.</p><p>You might think that once you’ve shared your insights, your job as a data analyst is done. In reality you want people not only hear what you’ve discovered but also act on your insights. This leads us to the final step — making people act on your data.</p><p><strong>Step 9. Make People Act on Your Data</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/519/1*dmJqRcBv3aQWcvnPWH8J6A.png" /></figure><p>Seeing my work have an impact and a chance to drive real change brings me the most satisfaction. So don’t let your hard work go to waste either.</p><ul><li>Work with the relevant teams to set clear action steps, timelines, and success metrics.</li><li>Monitor progress and ensure your recommendations are being implemented.</li><li>Communicate regularly with cross-functional teams to track the impact of your recommendations.</li></ul><p>I understand that this might feel like a lot right now, but please don’t worry. With practice, it will become easier, and before you know it, these steps will become second nature.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/522/1*FIUHYoNV9KSHxA7PK2OG0Q.png" /></figure><p>Good luck on your data analyst journey! You’re on the right track!</p><p><em>All images were created by the author using Canva.com.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8d45fb33e9c5" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-science/behind-the-scenes-of-a-successful-data-analytics-project-8d45fb33e9c5">Behind the Scenes of a Successful Data Analytics Project</a> was originally published in <a href="https://medium.com/data-science">TDS Archive</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Think you Know Excel? Take Your Analytics Skills to the Next Level with Power Query!]]></title>
            <link>https://medium.com/data-science/think-you-know-excel-take-your-analytics-skills-to-the-next-level-with-power-query-930e2267006e?source=rss-f44c0bb08544------2</link>
            <guid isPermaLink="false">https://medium.com/p/930e2267006e</guid>
            <category><![CDATA[data-analytics]]></category>
            <category><![CDATA[power-query]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[excel]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Ilona Hetsevich]]></dc:creator>
            <pubDate>Fri, 29 Nov 2024 16:09:48 GMT</pubDate>
            <atom:updated>2024-11-29T16:09:48.678Z</atom:updated>
            <content:encoded><![CDATA[<h4>5 practical use cases that prove Power Query is worth exploring.</h4><p>I have a confession to make: I’ve been living under a rock 🪨. Not literally, but how else can I explain not discovering <strong>Power Query</strong> in Excel until now?</p><p>Imagine realising that all those hours spent wrangling VLOOKUPs, nesting IFs, and battling messy data could’ve been replaced with a few simple clicks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/712/1*qzR1tyW9-KTPpuoAjFiCtw.png" /><figcaption>Image generated by the author</figcaption></figure><p>Power Query does everything Excel formulas could do — only faster, smarter, and way less frustrating. From merging datasets to effortless transformations and creating calculated columns, the possibilities are endless.</p><p>There are already tons of articles and videos out there with step-by-step guides to help you get started, so <strong>I won’t write another how-to</strong>. Instead, I’ll share the features that completely blew my mind with <strong>5 use cases </strong>— and hopefully inspire you to dive in and explore this powerful tool yourself. 🚀</p><p>To demonstrate the remarkable functionality, I will use a simple e-commerce dataset consisting of two CSV files: one with customer data and the other with transaction data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/221/1*kkRGjLEAPNyilKjeMHq71A.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/651/1*0cuhWV5ShsnggosPaC0xcg.png" /><figcaption>These screens display samples of the first 10 rows from each CSV file (generated by the author)</figcaption></figure><h3>Use Case 1: Merging Datasets (A Smarter Alternative to VLOOKUP)</h3><p>When it comes to combining data, we all turn to VLOOKUP. But let’s face it — VLOOKUP has its limitations.</p><p><strong>Why Power Query is better than VLOOKUP✨:</strong></p><ul><li>You can join data from multiple columns in one step (no need for repeated lookups).</li><li>Your lookup column doesn’t have to be on the left — it works no matter where it’s located.</li><li>It supports different join types (left, right, full, inner, anti), giving you more control over how data is combined.</li><li>Unlike VLOOKUP, Power Query doesn’t limit you to joining on just one column.</li><li>You can even handle messy datasets with typos using fuzzy matching!</li></ul><p><strong>How to merge datasets in Power Query 🔗</strong>:</p><ol><li>While on the <strong>Home </strong>tab select <strong>“Merge Queries”.</strong></li><li>Choose the two datasets (or files) you want to combine.</li><li>Select the column to join on and specify the type of join (e.g., left join to keep all rows from one table, or inner join to match only overlapping rows).</li><li>Once merged, select the columns you want to include in your final dataset.</li><li>Load the transformed data back into Excel.</li></ol><p>Done in seconds! ⏱️</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jhFeSRnda6Jdan4-9tiTQw.gif" /><figcaption>This gif explains the process of merging two files using Power Query (generated by the author)</figcaption></figure><h3>Use Case 2: Column Transformations</h3><p>Power Query makes it super easy to transform your data. From quick calculations to date handling and creating bins, it does in seconds what could take minutes (or more) in Excel.</p><p>Let me show you how with a few examples. 👀</p><h4>Perform Quick Calculations ➕➖➗✖️</h4><p>Let’s say you need to convert prices from USD to Euro and calculate the total value of your sales. In Power Query, you can:</p><ol><li>Go to the <strong>“Transform”</strong> tab and, under “<strong>Standard”</strong>, select <strong>“Multiply.”</strong></li><li>Multiply all the prices by 0.95 to convert the currencies.</li><li>Add a new column to calculate total sales by multiplying the price by the quantity. To do this, go to the <strong>“Add Column”</strong> tab and select <strong>“Custom Column.”</strong></li></ol><p>All of this happens in just a few clicks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*O3XZwjP8N-GGPrftgaV9dQ.gif" /><figcaption>This gif explains the process of performing quick calculations using Power Query (generated by the author)</figcaption></figure><h4>Modify Dates with Ease 📅</h4><p>Power Query makes working with dates straightforward too. You can quickly extract the month name or display only the first three letters (e.g., Jan, Feb, Mar) for a cleaner look using built-in functionality.</p><ol><li>Select the column that is in Date format.</li><li>Go to the <strong>“Add Column”</strong> tab, and under <strong>“Date”,</strong> select<strong> “Month”</strong> followed by <strong>“Name of the Month”</strong>.</li><li>Select the new column that contains the name of the month. Under <strong>“Extract”</strong>, choose <strong>“First Characters”</strong>, and in the pop-up window, specify how many characters you want to extract (3 in this case).</li></ol><p>No complicated formulas needed!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BGHm1tiBeFw0atEhdbXTvQ.gif" /><figcaption>This gif explains the process of modifying dates using Power Query (generated by the author)</figcaption></figure><h4>Create Bins Without Complex Formulas 📥</h4><p>I don’t know about you, but I always forget the IF formula syntax for creating bins in Excel, and it can get pretty long if you have multiple ranges. But with Power Query, it’s much easier.</p><ul><li><strong>If you’re using Windows</strong>, you’ve got the <strong>“Columns from Examples”</strong> feature. All you need to do is type one example of the bin range (e.g., “41–50”), and Power Query will automatically fill in the rest for you, saving you time.</li><li><strong>For those of us on Mac </strong>(like me), unfortunately, this feature isn’t available. But don’t worry, there’s a workaround! You can still create bins using the <strong>“Conditional Column”</strong> feature by setting logic rules to categorize your data into ranges.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*haI8f7bEoc5_rlgefhGWzg.gif" /><figcaption>This gif explains the process of creating bins using the “Conditional Column” feature (generated by the author)</figcaption></figure><h3>Use Case 3: Filling in Missing Data 🔎</h3><p>We’ve all worked with datasets with missing values — whether it is because of incomplete entries or data discrepancies. In most cases, you don’t want to leave this gaps, instead you want to fill them in. This is where Power Query becomes particularly useful.</p><p>Let’s say we have missing values in the “Price per Unit” column for the “Beauty” category, and we want to replace those missing values with the average price for that category. Here’s how you can do it in a few simple steps:</p><ol><li>Filter the “Product Category” column to show only “Beauty” entries.</li><li>With the “Price per Unit” column selected, check the average price for the Beauty category in the statistics section at the bottom of the screen.</li><li>Remove the filter on the “Product Category” column to show all categories again.</li><li>Select the “Price per Unit” column again, go to the <strong>“Transform”</strong> tab, and click on <strong>“Replace Values”</strong>.</li><li>In the dialog box, enter “null” for the value you want to replace and the average price for the Beauty category in the “Replace With” field.</li></ol><p>And just like that, you’ve filled in the missing data with the average value — all in a few clicks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*owZIQPpmbgERrZzdcnSttg.gif" /><figcaption>This gif explains the process of filling the missing values using Power Query (generated by the author)</figcaption></figure><h3>Use Case 4: Transforming Data for Better Analysis 🛠️</h3><p>Power Query is great for transforming your data into a format to match the needs of your analysis.</p><p>For example, if you want to summarise total sales per month and see the trend over time. You can use the “<strong>Group By”</strong> and “<strong>Transpose”</strong> functions.</p><p>Here’s how to do it in just 4 steps:</p><ol><li>Select the “Month” column to group your data by month.</li><li>In the Transform tab, click on <strong>“Group By”</strong>.</li><li>Add a new aggregation for “Total Value Euro” and select Sum to calculate the total sales for each month.</li><li>Finally, click on <strong>“Transpose”</strong> to switch rows and columns, turning months into column headers.</li></ol><p>Once you’re done, load the data back into Excel and build your line chart to visualize the sales trends over time!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sRpHCcYw3h6r0lDHZ6sOvQ.gif" /><figcaption>This gif explains the process of using Group By and Transpose functions (generated by the author)</figcaption></figure><h3>Use Case 5. Custom Columns with M Formula Language 🪄</h3><p>M formula language lets you go beyond the typical Power Query transformations, allowing for more advanced calculations and logic. It’s perfect when you need to create custom solutions for your data.</p><p>For example, let’s say the months in your sales data aren’t sorted correctly. Instead of manually rearranging them, you can use <strong>M formulas </strong>to assign a numerical value to each month, then sort them in the right order.</p><ol><li>Go to <strong>“Transform”</strong> and select <strong>“Custom column” </strong>option.</li><li>In the pop up window type the logic using “if-else if” logic to assign a numerical value to each month.</li><li>Sort by the new column in ASC order.</li><li>You can even remove the “Month Order” column once the sorting is done.</li></ol><p>After doing that, your months will be in the right order. 🏆🏆🏆</p><pre>if [Month short] = &quot;Jan&quot; then 1<br>else if [Month short] = &quot;Feb&quot; then 2<br>else if [Month short] = &quot;Mar&quot; then 3<br>else if [Month short] = &quot;Apr&quot; then 4<br>else if [Month short] = &quot;May&quot; then 5<br>else if [Month short] = &quot;Jun&quot; then 6<br>else if [Month short] = &quot;Jul&quot; then 7<br>else if [Month short] = &quot;Aug&quot; then 8<br>else if [Month short] = &quot;Sep&quot; then 9<br>else if [Month short] = &quot;Oct&quot; then 10<br>else if [Month short] = &quot;Nov&quot; then 11<br>else if [Month short] = &quot;Dec&quot; then 12<br>else null</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*O9P3xV7Gwbfrjb9Rq_eCeg.gif" /><figcaption>This gif explains the process of using M formula language in Power Query (generated by the author)</figcaption></figure><h3>Last but not the least…</h3><p>Power Query keeps track of every change in an applied steps log 💾, so if you want to go back and modify/undo anything, it’s super easy.</p><h4>Feeling inspired? 🧙‍♂️</h4><p>I hope you’re now as excited to try these features as I was! If I were you, I’d be jumping into Power Query right away.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=930e2267006e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-science/think-you-know-excel-take-your-analytics-skills-to-the-next-level-with-power-query-930e2267006e">Think you Know Excel? Take Your Analytics Skills to the Next Level with Power Query!</a> was originally published in <a href="https://medium.com/data-science">TDS Archive</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>