<?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[Enjoy Engineering - Medium]]></title>
        <description><![CDATA[Bringing joy to the world of startups, product, and tech. - Medium]]></description>
        <link>https://medium.com/enjoy-engineering?source=rss----282d70e4719d---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>Enjoy Engineering - Medium</title>
            <link>https://medium.com/enjoy-engineering?source=rss----282d70e4719d---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 14:08:23 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/enjoy-engineering" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[How Strategic Leaders Scale]]></title>
            <link>https://medium.com/enjoy-engineering/how-strategic-leaders-scale-aeee1137b511?source=rss----282d70e4719d---4</link>
            <guid isPermaLink="false">https://medium.com/p/aeee1137b511</guid>
            <category><![CDATA[management]]></category>
            <category><![CDATA[product]]></category>
            <category><![CDATA[leadership]]></category>
            <category><![CDATA[engineering]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Philippe Huibonhoa]]></dc:creator>
            <pubDate>Wed, 31 Mar 2021 00:16:07 GMT</pubDate>
            <atom:updated>2021-07-01T00:24:59.185Z</atom:updated>
            <content:encoded><![CDATA[<h3>How Strategic Leaders Scale —Fostering a Problems Not Solutions Culture</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*znnEeQJiTAQvlZtvQAXM1A.png" /><figcaption>What you celebrate as a leader determines the real values and culture of your organization.</figcaption></figure><p>The allure of dreaming up the next big thing or being “the one” who thought up “the answer” is often disproportionately celebrated. We know, especially in a start up environment, disciplined and thoughtful execution is as valuable, if not more, than a mere idea, but I will admit celebrating excellent execution is a challenging cultural problem. After all, it’s exciting to dream up new ideas, to brainstorm, to problem solve. It’s energizing to throw out ideas and easy to get caught up imagining the tremendous impact they can have. It’s part of the allure and the fun of joining a startup, isn’t it?</p><p>This “idea” culture misses what makes leaders truly great, though. The strength of a leader is fundamentally based on their relationship with their teams- and that relationship should not be one of commandeering authority, but one rooted with deep trust, support, honesty, and even sacrifice. The strongest leaders make everyone around them better, happier, and more impactful; by serving their team over themselves, they amplify the impact of the whole rather than any one individual.</p><p>Strong leadership is about setting a clear and strategic destination, but adopting a <strong><em>Problems Not Solutions</em></strong> mentality for the journey. Simply put, don’t tell teams what to build, instead give them ownership over the problem you are trying to solve. Trying to increase conversion? Don’t tell your team to swap to a one page checkout, tell them to increase conversion and give them the room and support to brainstorm, experiment, and iterate. At first it will be uncomfortable relinquishing so much control, but ultimately it is liberating to entrust your team to solve problems and frees you to focus on supporting them and to ensure the right teams are attacking the right problems at the right times.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*m7u_kJK422ERtH-5Ntnauw.png" /><figcaption>Leadership through problems not solutions is uncomfortable, at first, but eventually liberating. Leadership through “helicopter management” is comfortable but ultimately unscalable and inevitably debilitating.</figcaption></figure><h3>Empowering teams through ownership</h3><p>At first, pushing problems down the chain sounds like something a horrible leader does, but it is actually quite the opposite. It is a testament of trust that you did not hire someone just to be a pair of hands.</p><p>I’ve known my fair share of leaders who wanted employees to be exact clones of themselves, where success is defined as doing exactly as the leader would. On the surface this seems like it an effective tactic, after all, a leader must be smart and skilled to have been promoted to that level, so multiplying their reach should multiply their impact. In reality those who employ this style of “helicopter management”, are people who excelled as individual contributors- and were promoted to leadership for that success- but never adjusted their mindset, tactics, and problem-space for the leadership domain. Effectively they’re applying an individual contributor mentality and expecting it to scale and solve leadership problems. Especially in the start up world, this is wholly ineffective as it diminishes an employee’s sense of ownership and accountability, and suffocates any creativity and risk-taking required for innovation.</p><p>It also does not bode well for the leader. Rather than focusing on supporting their team, organizational efficiency, long term strategy, and everything else only a leader can do, they instead spend their time learning, checking, second-guessing, and correcting every decision and action their team makes. As the team inevitably grows, the leader must stretch like a rubber band being pulled to wrap around more and more before it finally snaps. It’s easy to see why, in these cases, leadership often devolves into stress-inducing micromanagement and burn out.</p><p>The corollary in the product world are those who see engineers as just builders- a pair of hands skilled enough to build, but still guided by another mind. Most engineers I know, however, don’t want to be just code monkeys. If you really dig in, engineers are not passionate about only building; they are passionate about problem solving; building is a means to solve problems.</p><p>In psychology and behavioral economics there is a very well studied phenomena, called the <a href="http://danariely.com/2012/09/20/real-world-endowment/">endowment effect</a>, that describes the tendency for people to ascribe more value to things merely because they own them. In much the same way, people have a tendency to be much more motivated, productive, and effective when working on ideas they own. Consider it the endowment effect applied to ideation. As a leader, you want to leverage that effect, not fight against it. This is precisely why empowering teams to solve problems, rather than just handing down solutions can be so powerful.</p><p>Give an engineer a problem that resonates with them, and you’ll see them produce some of the best work imaginable in very intense but sustainable ways. What’s more, engineers will bring new ideas and often better solutions to the table, because they know the details of how systems work.</p><p>At my previous start up, my team was close to launching a new inventory system for university bookstores. Imagine the incredible challenge of a new semester starting. Bookstore clerks would have to order, receive, and stock hundreds of textbooks within just a few days. As the stampede of students come in, they would then need to pick and pack the needed textbooks for each student.</p><p>We had asked Will, one of our talented engineers, to build a way to print ISBN (think SKUs for books) labels for the shelves to aid clerks in finding the correct textbooks as they pick and pack. Label printing is hardly the most exciting work, but of course Will, being a team player, got it done. The 13 digit ISBNs would print in bulk, in a good clear font, that clerks could affix to the shelves. A job well done.</p><p>Fast-forward to our in-office dry run. This was the launch of a major new service for our company, and we had insisted we need to do a dry run where engineers and product managers would play the role of bookstore clerks, and the rest of the office would be students picking up books from our conference-room-turned-bookstore. Will was running the pick and pack station. We had injected some problems (hard-to-scan barcodes, mislabeled books), but none of these were even a hiccup for Will, after all, these were problems we were already aware of and thus our processes and systems already guarded for it. But what Will did find was it was very difficult to find the right textbook even with the labels he had generated.</p><p>13 digit labels along shelving is just not easily scannable for a human. In fact by the end of the first day, we had discovered (and had a good laugh) that Will had not once, but twice, packed the wrong textbook.</p><p>When we reconvened the next day, however, Will excitedly shared he spent the night coding up a new solution for shelve labels: he built out labels that stuck out of shelves showing the author last name and the last 4 digits of the ISBN in BIG, BOLD type. That simple change was one of our customers’ favorite features, to the point that many bookstore managers would laugh and ask why they hadn’t been labeling their shelves like this all these years. And, of course, they’d ask how we came up with the scheme, to which we always happily shared Will’s story.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uXNH3GWiw0tZH8qH_OWQdQ.png" /><figcaption>In trench-warfare prioritization, everyone digs in, and no one wins.</figcaption></figure><h3>Impartial prioritization</h3><p>While the natural affinity to get excited about our own ideas can be really powerful, it can also lead to a very common pitfall: pet project prioritization. When people talk solutions, it is hard to stay impartial, because it is so easy for people to fall in love with their own ideas. Most of us have scars from these trench-warfare style prioritization meetings. Every inch is a grueling battle, as each person digs in, and volleys opinions at the other. In these types of meetings, there are no winners, just wasted time, frustrated employees, and bad decisions.</p><p>When you limit meetings to discussing problems, not solutions, the conversation is dramatically different. The meeting cannot spin off the rails with people just throwing out ideas (i.e. opinions on solutions) because there are no solutions for people to attach to. Moreover, scoping the priority and impact of problems lends itself to a more tangible and objective discussion- whereas the impact of a solution comes down to guessing the future, the impact of a problem is usually historical and computable.</p><p>Meetings are rarely something to smile over, but it’s exactly these problem-oriented meetings that excite me; ones focused on understanding the problems that face us and discussions not about how to solve them, but rather which ones our teams should solve first. In fact, one of my favorite measures of progress at any company is not checking that we have fewer problems to solve, rather it is checking to make sure we’re facing newer and better problems. (If you ever think you’re out of problems to solve, you probably aren’t looking hard enough or you’ve cornered yourself into a position ripe for disruption.)</p><h3>Outcomes over output</h3><p>Perhaps the most important consequence of a <strong><em>Problems Not Solutions </em></strong>mindset is that it changes the definition of success. Going back to the example of increasing conversion: if I go to my team and ask them to build a new checkout flow, their definition of success is, “did we build the new flow?”. If, however, I ask them to increase conversion in our checkout by 10% in the next 3 months, their definition success is now one and the same as mine.</p><p>The entire dynamic and approach changes. It is not about blocking enough time to build the new flow and accounting for every detail in the mock, instead it is about understanding what is impacting conversion, trying new ideas as quickly as possible, and leaving time to gather results and iterate. Their mindset is oriented around outcomes not just output, and this is absolutely key to scaling your impact as a leader. When your teams are obsessed and empowered to make an impact on the business, your job is all about making sure they are impacting the right areas and that they are setup to succeed. Who better to devise solutions than the executing teams, anyway? They know the systems, the costs, and the possibilities the best.</p><h3>Key takeaways</h3><p><strong><em>Problems Not Solutions</em></strong> is a powerful principle that helps strategic leaders scale. Adopting this principle…</p><ul><li>leaves solutioning to those closest to the solutions</li><li>drives accountability with teams and forces an outcomes over output approach</li><li>allows teams to build their ideas which leads to highly motivated teams, incredible innovation, and intense, sustainable work</li><li>means prioritization is no longer just a game of opinions or pet projects. Prioritizing the problem, forces you to answer: What problems are important to solve now and why?</li><li>lets leaders focus on the strategic space: making sure the right teams are attacking the right problems at the right time</li></ul><p>As a leader, I spend a lot of time thinking about how to create an environment that fosters sustainable productivity, impactful work, and happy employees. Through that lens I have formed many tenets over the years, but there’s none that I value more than <strong><em>Problems Not Solutions.</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/126/1*7B4qiDOhy_V8VACIQzU85Q.png" /></figure><p><strong>Written by Philippe Huibonhoa</strong>, VP of Product and Engineering at <a href="https://www.enjoy.com/en-us/">Enjoy</a>.</p><p>Do you want a hand in shaping the future of ecommerce? We’re hiring, <a href="https://www.enjoy.com/en-us/pages/careers">check us out</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=aeee1137b511" width="1" height="1" alt=""><hr><p><a href="https://medium.com/enjoy-engineering/how-strategic-leaders-scale-aeee1137b511">How Strategic Leaders Scale</a> was originally published in <a href="https://medium.com/enjoy-engineering">Enjoy Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Ruby and Rails can be as fast as you need — if you execute in parallel]]></title>
            <link>https://medium.com/enjoy-engineering/ruby-and-rails-can-be-as-fast-as-you-need-if-you-execute-in-parallel-632f2dff5429?source=rss----282d70e4719d---4</link>
            <guid isPermaLink="false">https://medium.com/p/632f2dff5429</guid>
            <category><![CDATA[ruby-on-rails]]></category>
            <category><![CDATA[ruby]]></category>
            <category><![CDATA[performance]]></category>
            <dc:creator><![CDATA[Michael Pearce]]></dc:creator>
            <pubDate>Tue, 21 May 2019 18:40:31 GMT</pubDate>
            <atom:updated>2019-07-18T00:56:34.772Z</atom:updated>
            <content:encoded><![CDATA[<h3>Ruby and Rails can be as fast as you need — if you execute in parallel</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kqffUcPtILz5dEvqjxZkfg.jpeg" /></figure><p>At Enjoy, we love Ruby and Rails for being vital pieces of a productive environment for developers. With every new Ruby or Rails version, developers can do more and do more easily.</p><p>What we don’t love Ruby and Rails for is their performance. Ruby is not known as the fastest language and as Rails continues to grow in features, it also sometimes falls behind in speed.</p><h3>We need to be fast, can we also keep it simple?</h3><p>Some of our web services at Enjoy have very strict performance requirements. If we executed everything serially in the web request thread, we wouldn’t be able to meet our SLAs. We need to execute code in parallel and we want to do it in as simple and fast a way as possible.</p><p>There are a few common options to performing code in parallel. One is to execute the code in threads which limits the parallelism to the number of cores and is often a non-starter for any GIL-using Ruby implementations. Another option is to extract certain parts of your application into services building a Service Oriented Architecture. This has the extra cost of decoupling the “client” and “server” and adds a new requirement of making sure both stay in sync.</p><p>A third method which we use at Enjoy for our most performance critical APIs is to partition work among a pool of waiting workers using our <a href="https://github.com/EnjoyTech/parallel_workforce">Parallel Workforce gem</a>. With this gem, the calling Rails web request thread splits a task into sub-tasks and enqueues jobs for Sidekiq workers dedicated solely to these sub-tasks. These workers do their piece of the task while the calling thread waits until each worker has responded with its result. The calling thread then aggregates the response and continues to execute as usual. Using this pattern, Enjoy parallelizes up to 20 different jobs at the same time for some requests. And it’s not hard to do.</p><h3>A simple example</h3><p>Using the gem is easy. We’re going to create some cryptographic hashes of passwords in workers using BCrypt, a hash algorithm designed to use a large amount of computing resources and be “slow.” We’ll then verify the generated hash against the password.</p><p>First, we’ll create an actor class with a keyword arguments initializer that accepts our password and a perform method that calculates the password hash.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/2addba3cd8a13dccd2ae931003ca2207/href">https://medium.com/media/2addba3cd8a13dccd2ae931003ca2207/href</a></iframe><p>We can then invoke the actors in parallel using ParallelWorkforce.perform_all with the actor class and the arguments for each invocation. Once we have the resulting password_hashes, we then validate they are correct.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/2206d0fc687c412f78ba991fca0d92de/href">https://medium.com/media/2206d0fc687c412f78ba991fca0d92de/href</a></iframe><p>When executing in parallel, the 3 password hashes can be calculated in around 1/3 the time as performing the same actions serially.</p><h3>How Parallel Workforce forces work in parallel</h3><p>Parallel Workforce makes use of either Sidekiq Workers (recommended) or Rails Jobs to do the actual work, but it’s Redis that ties everything together. Redis has become a key component of most web application stacks. It’s light weight, fast, and has some really neat features. One of those features is vital for our parallel workers — <a href="https://redis.io/commands/blpop">blpop</a>. The command causes a thread to wait until a value is inserted at a specific key — exactly what we need our calling thread to do while waiting for the workers to finish!</p><p>An outline and description of the execution are below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*y-8i2y4LSAuY7NUKUreukA.png" /><figcaption>Parallel Workforce job execution</figcaption></figure><p>The calling thread is responsible for splitting work and enqueuing a job for each work partition. Along with the class that performs the work and the arguments needed, each job also gets a Redis key of where to store its resulting computation with <a href="https://redis.io/commands/rpush">rpush</a>. Once jobs have been enqueued, the calling thread simply sits back and waits (i.e. blocks) using blpop on the result key until each of the workers has invoked rpush with its result.</p><p>The implementation is simple and works very well in practice.</p><h3>More to love with Parallel Workforce</h3><p>There are a few things to love about using the Parallel Workforce gem.</p><p>It’s easy to implement. If you have code you want parallelized and have a worker pool set up, you can easily refactor to make use of the gem. The process is very similar to any other refactoring to extract functionality into a separate class.</p><p>Deploys are easy. At Enjoy, we have zero-downtime deploys and we deploy often. For every new deployment, we first deploy our Rails app soon followed by the Sidekiq workers. When an executing Rails request web thread determines that the Sidekiq workers are running a different revision of code due to our workers being deployed a bit after our web server, the web thread automatically executes the job serially in the Rails web thread. Once the Sidekiq workers have restarted and have the same revision as the Rails web thread, parallel execution begins again.</p><p>Testing is seamless. We can integration test everything end-to-end with little more difficulty than if we weren’t doing any kind of parallel execution at all. There is no need to mock and stub web service APIs because all of the code is in the same repository and all of it can be executed in a test.</p><p>It’s fast! Redis has very low latency for the blpop and rpush commands. As soon as all your last worker has finished, the calling thread is ready to go!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/126/1*7B4qiDOhy_V8VACIQzU85Q.png" /><figcaption>Enjoy is reinventing retail in a mobile world.</figcaption></figure><p><strong>Written by Michael Pearce</strong>, Software Engineer at <a href="https://www.enjoy.com/en-us/">Enjoy</a>.</p><p>Do you want a hand in shaping the future of e-commerce? We’re hiring, <a href="https://www.enjoy.com/en-us/pages/careers">check us out</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=632f2dff5429" width="1" height="1" alt=""><hr><p><a href="https://medium.com/enjoy-engineering/ruby-and-rails-can-be-as-fast-as-you-need-if-you-execute-in-parallel-632f2dff5429">Ruby and Rails can be as fast as you need — if you execute in parallel</a> was originally published in <a href="https://medium.com/enjoy-engineering">Enjoy Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Graphql and make iOS Extensions re-usable]]></title>
            <link>https://medium.com/enjoy-engineering/graphql-and-make-ios-extensions-re-usable-2804db57c81b?source=rss----282d70e4719d---4</link>
            <guid isPermaLink="false">https://medium.com/p/2804db57c81b</guid>
            <category><![CDATA[graphql]]></category>
            <category><![CDATA[swift]]></category>
            <dc:creator><![CDATA[Matthew Cai]]></dc:creator>
            <pubDate>Fri, 10 May 2019 20:15:41 GMT</pubDate>
            <atom:updated>2019-05-10T20:15:41.482Z</atom:updated>
            <content:encoded><![CDATA[<h3>Graphql and Reusable Swift Extensions</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*CSgWoQAHMOzWmNR9zWv22w.png" /></figure><h3>Overview</h3><p>In the code sample below, the query is requesting information on a car.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/c4b2e2a997f0ef601c8404573c62b19a/href">https://medium.com/media/c4b2e2a997f0ef601c8404573c62b19a/href</a></iframe><p>Everything works completely fine the way it is currently structured.</p><h3>Reusable</h3><p>However, when fragments and extension are included everything gets a little more complex.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/096d5ce671d63bf44886bafd01713374/href">https://medium.com/media/096d5ce671d63bf44886bafd01713374/href</a></iframe><p>The above code creates a fragment that can be reused between different queries.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/08f9d18ade3f72031dd36bd2a310a0e4/href">https://medium.com/media/08f9d18ade3f72031dd36bd2a310a0e4/href</a></iframe><p>Extending on the useful feature of fragments, adding extensions on the built models creates more flexibility.</p><h3>Problem</h3><p>So what are the problems with the presented examples?</p><p>Currently, we can structure our original query like this:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/37f300c13bd90a41cba880d998212ad8/href">https://medium.com/media/37f300c13bd90a41cba880d998212ad8/href</a></iframe><p>What happens when we want a query that does not include all the information that the fragment provides?</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/f800eaec0cf0c059b58f00b4c59e584b/href">https://medium.com/media/f800eaec0cf0c059b58f00b4c59e584b/href</a></iframe><p>The above query(expanded version of the fragment) is missing <em>size</em> and <em>isUsed</em> that is required in the fragment. As a result, we can no longer use the extension applied to the fragment that we created earlier.</p><h4>How do we resolve this?</h4><p>Do we create multiple fragments and implement the same extension onto them to solve this issue or is there a better way to structure our fragments?</p><h3>Solution</h3><p>We can accomplish the same result by using a mix of protocols and generics.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/8c4083d4fe6c91b67cf7bc87c3eab253/href">https://medium.com/media/8c4083d4fe6c91b67cf7bc87c3eab253/href</a></iframe><blockquote><em>CarWheelsDetailInfo</em></blockquote><p>In the code sample above, we start by creating a protocol <em>CarWheelDetailInfo</em> on the attribute we plan to use in our extension. In our case, the attribute is the <em>cost.</em></p><blockquote><em>CarWheelsDetailInfo</em></blockquote><p>Next, we create a protocol <em>CarWheelsDetailInfo </em>that will include <em>CarWheelDetailInfo </em>as a generic.</p><p>Now we can apply <em>CarWheelsDetailInfo </em>on anything that includes the wheel’s cost in the query or fragment. The computed property <em>wheelPrice </em>will always be included with the protocol.</p><h3>Summary</h3><p>Use generics and protocols as a way to make extensions reusable.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/126/1*obtNAUpkBehVzCy5sFwDqw.png" /><figcaption>Enjoy is reinventing retail in a mobile world.</figcaption></figure><p><strong>Written by Matthew Cai</strong>, iOS Engineer at <a href="https://www.enjoy.com/en-us/">Enjoy</a>.</p><p>Do you want a hand in shaping the future of ecommerce? We’re hiring, <a href="https://www.enjoy.com/en-us/pages/careers">check us out</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2804db57c81b" width="1" height="1" alt=""><hr><p><a href="https://medium.com/enjoy-engineering/graphql-and-make-ios-extensions-re-usable-2804db57c81b">Graphql and make iOS Extensions re-usable</a> was originally published in <a href="https://medium.com/enjoy-engineering">Enjoy Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Don’t Dilute Your Product!]]></title>
            <link>https://medium.com/enjoy-engineering/don-t-dilute-your-product-bcb5b6f0757e?source=rss----282d70e4719d---4</link>
            <guid isPermaLink="false">https://medium.com/p/bcb5b6f0757e</guid>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Philippe Huibonhoa]]></dc:creator>
            <pubDate>Tue, 30 Apr 2019 21:14:43 GMT</pubDate>
            <atom:updated>2020-09-29T00:41:49.899Z</atom:updated>
            <content:encoded><![CDATA[<h4>The secret behind products that pack a punch 👊</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZGnxdZKU40vrrZoWtjF_CA.png" /></figure><p><strong>Product Management Bugs</strong></p><p>The definition of a bug is: <em>“anything that causes the system to produce an unexpected result”</em>. Strangely enough, we tend to view bugs solely as an engineering construct, when the product management domain is just as susceptible, if not more.</p><p>What is a product management bug? If product management is, at its core, about ensuring you build the right things, product management bugs are cases where you thought a feature would do one thing, but it either fails at its job or does something else entirely.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/550/1*zS8mJ33dvXXkbviBLp63ww.jpeg" /></figure><p>Years ago, I was working on a project where it was clear visitors didn’t understand our value-proposition, and thus were not converting. I reached out to a few customers, and found that with a few tweaks to how I positioned what we do, customers ‘got it’ and were more than ready to give us a try. Armed with this new messaging, we crafted a beautiful video that everyone on the team agreed was on-point, on-tone, and on-brand. Homerun. I put up the video, top-and-center, where it wouldn’t be missed and moved on to the next problem, after giving myself a pat on the back for a job well done…</p><p>A few months go by and someone wants to move the video from the top of the page… crazy right? Didn’t they know how important this video was for our customers and our conversion rate? This sparked a lively debate. To make matters worse, we never actually said the video’s job was to increase conversion. Some thought, and argued, that the video couldn’t be moved because it was critical to our branding. Others argued that no one watched the video… while others, still, argued placement didn’t matter as long as it was somewhere on the page.</p><p>Finally, we agreed on an AB test; one where conversion would be the deciding factor. It was that day I learned our amazing video had actually been reducing conversion by 6%.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/550/1*Uahg1baHyrhW_ppDhWphEg.jpeg" /></figure><p>We’ve been conditioned to look out for bugs in the engineering world; after all, they are often glaring and customer-facing. When clicking on a button results in an error, that’s in everyone’s face. Product bugs are much more subtle, but that’s exactly what makes them so dangerous. When they’re around, there are no red flags, error notifications, or system alerts… things seem like they’re working fine. To sort out these bugs requires incredible discipline. You must both call out exactly what you think a feature will accomplish and measure the results after it’s been launched; you’re fighting both the fear of being wrong and the temptation to keep moving.</p><p>The problem is, if you ignore these bugs, that doesn’t make you any less wrong, it just means you don’t know you’re wrong and you can’t learn from it. Likewise, while it may seem like you have to slow down to catch these bugs, don’t be quick to confuse action with progress. Just because you’re working hard or launching a lot of features does not necessarily mean you’re making any progress, you could just be thrashing water.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vv3sP7U3zYVBKWnM4PjP0A.png" /><figcaption>Don’t confuse action with progress.</figcaption></figure><p><strong>Freeloading Features</strong></p><p>The cost of a feature is not just the time and resources needed to build it- that’s just the down payment. Every feature adds a drag coefficient to your execution speed, a drag that adds up surprisingly quick. It’s the difference between steering a speedboat and steering the Titanic.</p><p>When you add new features, you add complexity to your systems, clutter your UIs, and risk confusing your customers. From an engineering perspective it’s harder to find bugs, slower to iterate, and more work to refactor or upgrade code. Think about the difference in effort between remodeling a cozy cottage vs. remodeling a towering high-rise. From a design perspective, it’s harder to maintain consistency, clarity, and streamlined user flows. And finally, from a product perspective, features that are not useful to customers cause bloat and confusion.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uHWsN6ZjRBYIFS_NZ23DIQ.png" /></figure><p>If features are not doing their job, you’re paying these costs for no reason. You are sacrificing your ability to launch and nurture successful features in order to support product management bugs… and if you’re letting these bugs build up, all you’re doing is diluting your product.</p><p>Diluted products are those that do everything, but don’t do anything particularly well. Diluted products feel old, bloated, and cumbersome, because they’re unable to keep up, under the weight of supporting useful and useless features alike. Diluted products are easily disrupted by competitors who can focus on features that add value. Don’t dilute your product!</p><p><strong>Packing a Punch</strong></p><p>If you want your product to pack a punch, the strategy is simple: make sure every feature you are supporting is doing its job. The more you let PM bugs linger, the more you dilute your product, and the less effective you are, and the less effective your product comes across…</p><p><strong>1 Write down what you think the feature will do. </strong>This is the hardest, but the most important step- you actually have to define what the feature’s job is… down to the metric if applicable. (Don’t just say we think <em>Feature A</em> will increase conversion, put a number to it- <em>Feature A</em> will lift conversion by 25%). It’s ok to be wrong, you’ll learn and get better.</p><p>It is also not enough to just have an expectation, you should record it. Why?</p><ul><li><strong>Eliminate Confirmation Bias </strong>-<strong> </strong>Humans are wired to interpret results in a way that aligns with their preconceptions. Without writing down your expectation, you’re likely to unconsciously alter your expectation to match the results after the fact.</li><li><strong>As a Sanity Checkpoint</strong> - Recording your expectation is a good sanity check that you’re building the right feature and that everyone is on the same page. What problem are you trying to solve? Is there a better solution out there? Are you collecting the right data?</li><li><strong>Facilitate Learning </strong>- Stating your theory and checking if you’re right or wrong will help you calibrate what works for your product; it’s a track-record you can review and learn from. Without it you end up making the same blind guesses release after release.</li></ul><p><strong>2 Invest properly. </strong>A quick way to get a feel of how much you should invest is to categorize where on the Kano Model a feature lies while factoring in how much validation you’ve done.</p><p>Is this a checkbox feature? Don’t over-invest. Is this your first go at the problem? Try an MVP. Is this an iteration on a key feature? Go big. Have customers put something on the line for you already (their email, their reputation through referrals, their money)? Go even bigger.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5HFRrmNSCZhW2sLHWHRB7g.jpeg" /><figcaption>Kano Model — What type of feature are you building?</figcaption></figure><p><strong>3 Build features with the necessary instrumentation to measure results. </strong>Know what metrics should be impacted and what metrics might tangentially be impacted. You can’t improve what you don’t measure; likewise, no feature is complete if the tools to measure it are not yet built.</p><p><strong>4 Release and learn.</strong> Releasing a feature IS NOT the finish line- that would be like giving birth to a child and calling it a job well done. <em>Are customers using the feature? Which customers? Is the feature doing its job?</em> This is the point when you start to get real feedback, so don’t just release and move on… unless your product strategy is a good, old fashion spray and pray.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eyKDMdzwsVNgkhWdz9DOFA.png" /><figcaption>Releasing a feature is not the finish line.</figcaption></figure><p><strong>5 Pivot, polish, or kill. </strong>Use what you learned to make an educated decision.</p><ul><li><strong>Pivot: </strong>The feature didn’t quite do its job, but gave you ideas for something that may do better…</li><li><strong>Polish: </strong>The feature did the trick, time to double down…</li><li><strong>Kill:</strong> The feature is a miss, now to take the time and discipline to remove it, otherwise you are just diluting your product…</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/126/1*Art5rXs8WttsiwcAhlQyNw.png" /><figcaption>Enjoy is reinventing retail in a mobile world.</figcaption></figure><p><strong>Written by Philippe Huibonhoa</strong>, Head of Software Engineering at <a href="https://www.enjoy.com/en-us/">Enjoy</a>.</p><p>Do you want a hand in shaping the future of ecommerce? We’re hiring, <a href="https://www.enjoy.com/en-us/pages/careers/all-jobs-positions">check us out</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bcb5b6f0757e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/enjoy-engineering/don-t-dilute-your-product-bcb5b6f0757e">Don’t Dilute Your Product!</a> was originally published in <a href="https://medium.com/enjoy-engineering">Enjoy Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>