<?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 Sergey Konovalov on Medium]]></title>
        <description><![CDATA[Stories by Sergey Konovalov on Medium]]></description>
        <link>https://medium.com/@sergeykonovalov?source=rss-5de21f672488------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*VLO6NNN9mB_gNGX_Z5giBA.jpeg</url>
            <title>Stories by Sergey Konovalov on Medium</title>
            <link>https://medium.com/@sergeykonovalov?source=rss-5de21f672488------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 20:20:18 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@sergeykonovalov/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[Technical Debt Does Not Exist]]></title>
            <link>https://medium.com/@sergeykonovalov/technical-debt-does-not-exist-f2982f0d6052?source=rss-5de21f672488------2</link>
            <guid isPermaLink="false">https://medium.com/p/f2982f0d6052</guid>
            <category><![CDATA[technical-debt]]></category>
            <category><![CDATA[unpopular-opinion]]></category>
            <category><![CDATA[engineering-management]]></category>
            <category><![CDATA[product-management]]></category>
            <dc:creator><![CDATA[Sergey Konovalov]]></dc:creator>
            <pubDate>Sat, 25 Jan 2020 21:41:26 GMT</pubDate>
            <atom:updated>2020-01-25T21:41:26.868Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CASl1JKR5_cFlNf6-qU5WQ.jpeg" /></figure><p>There are over 180 millions of resources on the Internet, that mention the term “technical debt”, if I can trust Google. Surely there is no chance I’ve checked all of them, but I checked a lot to prove my assumptions:</p><ul><li>meaning of the term is very vague;</li><li>refers to some implied costs of prioritisation of delivery over perfect code;</li><li>hard to strictly define;</li><li>even harder to estimate.</li></ul><h3>The owls are not what they seem</h3><p>First of all, terminology: debt is something you are obliged to pay back. While in case of so-called technical debt we deal with probability, not a determined fact. Modified part might never be addressed again, or addressed years from now. It is therefore no way a debt, but a want.</p><p>Software engineering is the only industry adopted this term. It is true that some related industries such as design, marketing and management tend to adopt it as well, but from a practical perspective it is really hard to comprehend what scope is called design debt or organisational debt. There is naturally no debt in sports, medicine, communications, parenting and many other key functions that we run daily, because it’s hard or impossible to choose easier way together with some commitment to be resolved in the future.</p><p>Perfect code can be infinitely better, there is no answer what is baseline of good to refer to. Maybe 100% uptime is ideal, while 99,9% is sufficient for most. Maybe current design with expected failure in 3 years is exactly what is needed, because by that moment it will be a totally different thing. Even if you put it vice-versa and put consequence of so-called technical debt as harder ways to implement new features that it would be ideally. Still the question is how much harder?</p><p>If you do not know right now about even better way to solve problem by quickly injecting some solution or smartly doing something, will it become immediately your so-called technical debt as soon as you know it? Once you found it out, you learned. And all your beliefs about what was right before immediately become useless, outdated and not that smart. And haven’t we just created another so-called tech debt by finding new, much more efficient way to solve our problem that we did not know before?</p><p>Paying so-called technical debt back should make system better. But because there is a gap between the moment when so-called technical debt appears until people start working to reduce it, that means they do not take into account all learning that happens in-between.</p><p>Consequently, if they consider that changes and adjust their activity, it’s not paying debt anymore, it is further improvement. We do not fix, but work on new way providing value to our customers. As long as we discovered, validated and learned what is the issue for customers, otherwise we solve problems that do not exist.</p><p>Definition of technical debt states that it is always a change that has to be done. Assumption with term technical debt is that changes are improvements. Let us see how improvements are handled in automotive, the most efficient businesses where price of engineering changes is extremely high.</p><h3>Leaner means less waste</h3><p>Few things in the world can be harder than modifying production (<em>note it is literally production site in this context</em>) with thousands of people, thousands of components, equipment and maintenance processes. And it is not only about adding new activities, but also removing activities that add no value.</p><p>Success of automotive is largely based on principles of the Toyota Production System (TPS), every major automotive maker has own production system based on it. TPS in turn is largely based on the approach to change management, called continuous improvement.</p><p>The key concept here is that value-added operations is something that customers would pay for. Will your customers pay for suggested changes triggered by so-called technical debt?</p><p>There are 7 classical types of waste, that can be much easier mapped to software development, that you might think. Try to take a different look at your backlog to answer the questions below and decide if working on items will make your product leaner.</p><h4>Transportation</h4><ul><li>What will be the effect on the amount of sent data between existing systems?</li><li>Will we need separate migration of large data sets?</li><li>Will it cause delivery of data to new places or delivering from new places?</li><li>Do we need to use another server at different location?</li></ul><h4>Inventory</h4><ul><li>Will it require additional storage for operations?</li><li>Will it reduce or increase the amount of work in progress?</li><li>Will it require other related data to be prepared and kept for processing?</li></ul><h4>Motion</h4><ul><li>How would that influence probability of downtimes in our service?</li><li>How will that affect load of our hardware?</li><li>Is this a complication or simplification of logic inside of processing unit</li><li>Will it require us to run additional operations or processing?</li></ul><h4>Waiting</h4><ul><li>Would that cause delays for units down the pipeline?</li><li>Will units down the pipeline be capable to process our unit outputs?</li><li>How it will affect overall response time of our pipeline or process?</li></ul><h4>Over-production</h4><ul><li>Will it increase batch of data that we process?</li><li>Will it add extra data that is not used?</li><li>Is all of the data we provide is used?</li></ul><h4>Over-processing</h4><ul><li>Do we really need all of these features?</li><li>Are we happy with current uptime or we need to work to reduce time of outages?</li><li>Is the approximation sufficient or we really need higher precision?</li><li>Will it cause additional data formatting, serialisation, enrichment?</li></ul><h4>Defects</h4><ul><li>Is current quality good enough?</li><li>How frequent problem is?</li><li>How severe problem (does it affect main user scenarios and flows)?</li></ul><p>If you are producing so-called technical debt at high speed, you probably introducing features at high speed, and that means you are introducing potential waste at high speed. You should stop and think if all of them do yield value. Because if everything released would be valuable, you would iterate further on exploring areas related to identified value and would work on improving, or investing into your product.</p><p>That’s logical: if there is a value and market fit, there are resources. But if not, it is insane to think that working on perfection of things not needed will make it anyhow better. So-called technical debit therefore does not fight added complexity. It is added complexity.</p><h3>Value or learning (better both)</h3><p>Putting so-called technical debt into your backlog is one of the worst things one can do. First of all, this grow up your backlog and adding noise to draw your focus from things that matter. Don’t make your backlog list of sorry points.</p><p>Challenge the change: what is the worst consequence if we do this never? You should not separate improvement from your regular flow working with hypothesis and stories, it should be continuous improvement itself. Can be one minute thing, or can take whole sprint, does not matter if its value is superior to others.</p><p>Result that you strive should be value or <em>validated learning</em>. So it is absolutely acceptable to miss, if you earn competitive knowledge. Everything before validated learning is an assumption.</p><p>Maintain complexity as a product too: discover, validate and rank hypothesis of changes you want to introduce. If something has to be done, there should be facts and evidence for that (e.g. analytics, expertise, corridor test). This perspective makes you look wider and notice patterns, get ideas to simplify workflow, system thinking to generate high quality hypothesis on product improvement. Technical design is a product. Architecture is a combination of decision made to system that are hard or impossible to change. We shall be very precise when making such decisions.</p><p>Every so-called technical debt item shall have exact customer job it targets and measurable outcomes. What exactly will the customer get out of this change? No generic words “faster”, “better”, but give real numbers.</p><p>“Load speed will be 250 milliseconds faster, and that would make +0.575% to conversions as per stats we have, meaning $62 000 of revenue per year”. This is absolutely valid.</p><p>“This implementation will support 5 users, but at 6 it will fail”. It is not a debt, it is an acceptable solution in the current situation, if we have only 2 users. We might never have more than 3 users at all. It’s an honest and valuable trade.</p><p>Can you measure and visualise what modules in your system are most complicated and take most of changes? Can we name what are hot zones and allow zero tolerance to cruft there? Constraints facilitate us to find genuine better solutions, instead of assuming current solution as something temporary, as a way to work around, and postponing improvements for undetermined moment in future.</p><p>If delta between current solution and better one that we envision is not mandatory, it’s nothing more than a wish-list. Everything we deliver at any given moment should be the most valuable product and solution for the given situation. We should love and care about the current product, not its imaginary future. We do not need separate term “technical debt” for it, let’s call it product opportunity instead. Equal to other product opportunities that we have.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f2982f0d6052" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Lean Product System]]></title>
            <link>https://medium.com/@sergeykonovalov/lean-product-system-3e89b724674e?source=rss-5de21f672488------2</link>
            <guid isPermaLink="false">https://medium.com/p/3e89b724674e</guid>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[agile]]></category>
            <category><![CDATA[digital-product]]></category>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[business-models]]></category>
            <dc:creator><![CDATA[Sergey Konovalov]]></dc:creator>
            <pubDate>Mon, 24 Dec 2018 22:15:08 GMT</pubDate>
            <atom:updated>2018-12-24T22:15:08.842Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*9szGQum9Ip_qZ1wsBguJ0w.jpeg" /></figure><p>Being a product manager is tough, but rewarding fitness for brains. After about 7 years in operations management, business development and software engineering in companies all around the planet, I’ve come up with a simple yet solid set of craftsman tools serving me really well.</p><p>It’s probably not something radically new, there are <a href="https://www.product-frameworks.com/">tons of other good frameworks</a> available, made by really smart and experienced people. It is just very simple, and it just works for me.</p><p>Pilots go through control cards for every flight taxi, take-off and landing procedures. Being a very systematic kind of person, I found it very useful to bake all fragments of my knowledge and experience—and turned out to be not only my, but also of many colleagues and friends who significantly contributed — into a tasty cake I called <em>a product system</em>, full of control cards and checklists to save you from most prominent mistakes.</p><p>In our work uncertainty causes procrastination, so we can save tons of time, efforts and resources by just knowing how to approach and sole particular problem. So that we do not need to reinvent the wheel.</p><p>Shortcuts should not exclude common sense and thinking, probably not all of problems can be solved by this thing I am suggesting, and probably it is not fit for all organisations, not all teams, not all leaders, not all industries and domains. But if that would help to give you that 20% of tools to solve 80% of problem, I would be more than happy.</p><p>I’ve also noticed that IT industry and product management domain is now actively deploying tools and methodologies well known by large corporations for maybe two or three decades. Kanban, lean, total quality management, statistical process control, kaizen, waste elimination, six sigma, just in time and many others adopted from just engineering and management in automotive to become widely used by companies in finance, logistics, retail, agriculture, transport, construction and eventually in software development. It took time to rethink and apply models to fast world of startups, but eventually I am very happy to enjoy these tools in my today work. And that is reason why my work is <em>lean product system</em>.</p><p>Its main goal is to suggest instruments for any job that product manager tries to be done: from high-level understanding of business and its direction, to specific guidelines how exactly particular things have to be implemented.</p><h3>Stage 1: Define</h3><p>Everything starts with definition. Here you can still save money, not diving directly into risky and expensive adventures of actually building products. It may sound opposite to agile approach, where product is built iteratively as team gains expertise and become experts in some specific topic. However, this is exactly the opposite. The reason is that MVP is not necessarily something you need to build immediately. It is easier and much cheaper to utilise many other ways to do validated learning about market, customers and competitors, without writing a single line of code.</p><ul><li>Research global and industry specific trends</li><li>Analyse and benchmark competitors</li><li>Ideation</li><li>Discover opportunities (qualitative interviews)</li><li>Validate hypothesis (lean MVP process, validated learning)</li><li>Unit economy and forecasting</li><li>Writing clear specifications with mockups</li></ul><h3>Stage 2: Build</h3><p>This part is all about best practices, ceremonies and project management, tightly connected with coordination, updates, people management, and sharing common vision to all members. Main goal is to avoid dependencies, release with short iterations and feed product to market, because that usually means live real customer feedback and first money.</p><ul><li>Apply best practices about user interaction and design systems</li><li>Choose proper development methodology for team</li><li>Use tips and tricks to design engaging experience for future users (for example, awesome <a href="https://yukaichou.com/gamification-examples/octalysis-complete-gamification-framework/">Octalysis framework</a> by Yu-Kai Chou)</li><li>Integrate with useful analytics tools</li></ul><h3>Stage 3: Launch</h3><p>Focus on launch date and taking all necessary steps to make sure customers’ journeys go smoothly, hence money on marketing are well-spent. This covers not only preparation and actual moment of launch, but also following days or weeks. With lean approach, this is frequently an organic next step of product versions, delivered during building stage (you might have paying customers by that moment), so launch activities matter when you have already a working product with very basic features, and already need some materials for your sales.</p><ul><li>Writing attractive and effective marketing materials</li><li>Choosing and properly setting up best channels</li><li>Setting up sales process and handling of the leads</li><li>Training and education</li><li>Key metrics and data-driven decisions</li></ul><h3>Stage 4: Improve</h3><p>When you have validated value for customers, it is time to start building the company, overcoming growth problems, eliminating waste and work on further product improvements.</p><ul><li>Optimise funnels and conversions</li><li>Evolving standards and process maturity</li><li>Setup logistics, supply, delivery and other critical processes</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3e89b724674e" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>