<?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 Martin Fahy on Medium]]></title>
        <description><![CDATA[Stories by Martin Fahy on Medium]]></description>
        <link>https://medium.com/@mmartinfahy?source=rss-66fc8e112439------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*D9S27NySRuTTO2XF8hvUyA.jpeg</url>
            <title>Stories by Martin Fahy on Medium</title>
            <link>https://medium.com/@mmartinfahy?source=rss-66fc8e112439------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 10:31:27 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mmartinfahy/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[Understanding Timings in Distributed Systems]]></title>
            <link>https://klaviyo.tech/understanding-timings-in-distributed-systems-262fbc51441d?source=rss-66fc8e112439------2</link>
            <guid isPermaLink="false">https://medium.com/p/262fbc51441d</guid>
            <category><![CDATA[tracing]]></category>
            <category><![CDATA[distributed-systems]]></category>
            <category><![CDATA[jaeger]]></category>
            <category><![CDATA[metrics]]></category>
            <category><![CDATA[logging]]></category>
            <dc:creator><![CDATA[Martin Fahy]]></dc:creator>
            <pubDate>Mon, 18 Nov 2024 16:01:37 GMT</pubDate>
            <atom:updated>2024-11-18T16:01:37.215Z</atom:updated>
            <content:encoded><![CDATA[<h4>How we combined logging and tracing to make life much easier.</h4><p>Measuring the time spent between two points in a request’s lifecycle in a distributed system sounds simple enough, right? You just log the start and end times, look at a chart, and voilà, problem solved. But when a customer raises an issue about a specific delay, and you dive into your logs, charts, and metrics to diagnose it, you quickly realize that connecting the dots across disparate systems is anything but easy. Despite advanced tooling, observability graphs, dashboards, and well-structured <a href="https://klaviyo.tech/logging-klaviyo-how-we-log-10-billion-events-a-day-c05902b8b538">logging</a>, achieving a full picture of system behavior can still be challenging.</p><p>This problem recently became apparent for my team when a high-MRR customer reported delays in receiving specific SMS messages on their test numbers. Investigating these specific messages meant diving into our various observability tools to see if we could pinpoint the timing.</p><p>At Klaviyo, we rely on a range of tools to provide visibility into our system, including:</p><ul><li><strong>Splunk and Logstash</strong> for transforming and querying structured log data</li><li><strong>Grafana, Graphite, and StatsD</strong> for real-time metric aggregation and visualization</li><li><strong>Redshift</strong> for OLAP queries related to log events</li><li><strong>Jaeger and Clickhouse</strong> for tracing</li><li><strong>Sentry</strong> for exception tracking</li></ul><p>Each one of these tools are useful but have inherent limitations when it comes to providing a coherent picture of holistic execution.</p><h3>Splunk and Logstash</h3><h4><strong>Pro:</strong></h4><ul><li>Provide formatted content of structured log events with related unstructured data.</li></ul><h4><strong>Con:</strong></h4><ul><li>Distinct ordering of logs is crucial to understand how events relate to each other within the control flow.</li><li>There needs to be a specific identifier to associate these logs across domain boundaries.</li><li>Detailed time-related fields, including start, end, and duration timestamps, are needed to accurately capture the lifecycle of an event.</li></ul><h3>Grafana, Graphite, StatsD</h3><h4><strong>Pro:</strong></h4><ul><li>These tools provide an easy-to-visualize view of real-time events, making it straightforward to monitor system performance.</li></ul><h4><strong>Con:</strong></h4><ul><li>Cardinality becomes a concern with Graphite’s storage, particularly when it comes to rolling up aggregations. This limitation makes it impractical to create stat buckets based on unique identifiers, like request ids. Consequently, we often need to limit our stats to a higher level of granularity to maintain performance and manage storage efficiently.</li></ul><h3>Redshift</h3><h4><strong>Pro:</strong></h4><ul><li>Using SQL as the primary query language simplifies the process of performing joins and aggregations across various unstructured log sources and enables more versatile data analysis.</li></ul><h4><strong>Con:</strong></h4><ul><li>Executing queries can be slow, especially when dealing with a large volume of ingested log events.</li><li>Redshift provides only one guaranteed timestamp field, which indicates when a log was ingested, instead of granular timing stats like start, stop, and duration timings which need to be added explicitly.</li></ul><h3>Sentry</h3><h4><strong>Pro:</strong></h4><ul><li>Makes it easier to identify and troubleshoot code-level issues, bugs, and runtime errors in real time.</li><li>Captures detailed stack traces and data associated with errors which helps with providing context around errors.</li><li>Tracks user sessions and allows you to see how many users are affected by each error, which helps determine scope of an issue.</li></ul><h4><strong>Con:</strong></h4><ul><li>While effective for tracking exceptions, it does not inherently associate events across domain boundaries, making it challenging to connect events in a control flow.</li><li>Costs can rise quickly with high error volumes.</li><li>Sentry primarily focuses on exceptions rather than capturing start, end, or duration timings; not intended for non-exceptional event lifecycle tracking.</li></ul><p>We quickly realized that our existing tooling fell short in addressing what seemed like a straightforward question due to several limitations:</p><ol><li><strong>Association of events</strong>: Understanding how one event associates with another and how each fits within the overall control flow.</li><li><strong>Granularity of timings</strong>: Obtaining distinct timings for each event, including start time, end time, and duration (calculated as end time − start time).</li><li><strong>Sampling strategies</strong>: Acknowledging that not all events hold the same significance. We needed to prioritize “high-priority” events while potentially reducing instrumentation for less critical ones.</li></ol><p>For instance, when associating events, we encountered challenges with logs and events that were recorded in one system using specific identifiers relevant to that domain, but as these events crossed domain boundaries, the identifiers were not consistently propagated. For example, batch_id identifies a particular batch of SMS messages in one system, but a different system would log a transmission_id with the same batch, but not log the corresponding batch_id.</p><p>This discrepancy made it difficult to join events across systems since they lacked a common attribute for association. Even when we managed to join events, our logs often failed to capture essential timing information, such as start time, end time, and duration, for logs outside our domain, as well as the ordering needed to accurately associate the control flow of events.</p><h3>Jaeger for Distributed Tracing</h3><p>Recently, we introduced <a href="https://www.jaegertracing.io/">Jaeger</a> into our observability stack. What is a Jaeger? At a very high level it’s a tool that maps the flow of requests and data as they traverse a distributed system.</p><p>Jaeger gives you the following in a trace:</p><ol><li>A unique trace id.</li><li>A distinct start and stop time for the event.</li><li>A sequential ordering of events.</li><li>A visual breakdown of per method execution.</li><li>The ability to capture the input and output of a function.</li><li>Context propagation meaning that if others implement tracing in their specific domain you get insight into the execution of that system for free.</li></ol><p>Simply put, in Python, we add a decorator at the start of a workload and Jaeger automatically traces the entire execution flow.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BpkMGGBQDVayy-uVqZPo2g.png" /><figcaption><em>Jaeger UI showing the request path of processing a transmission batch.</em></figcaption></figure><p>Unfortunately, utilizing Jaeger does not come without its challenges. The spans generated by Jaeger can become quite large, making it impractical to trace every request.</p><p>As a result, our sample rate for traces is set to a default of <em>0.01%</em> of all traffic. While this helps manage performance and storage, it comes with the downside of providing only a high level understanding of overall system performance. Consequently, we are often unable to track specific events, such as those with a particular customer.</p><h3>Enhancing Observability with Associated Logs and Traces</h3><p>Recently, our team overcame several limitations by implementing a few strategic changes.</p><p>First, we integrated Jaeger trace ids into our Splunk logs, allowing us to easily identify Jaeger events from any Splunk log entry. By associating these logs with Jaeger spans, we are able to efficiently look up the corresponding trace in Jaeger:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DSF9dPJ4DGyov4iAW_AGUQ.png" /><figcaption><em>By logging trace_id, we can easily move from a log entry in Splunk to the Jaeger trace.</em></figcaption></figure><p>Incorporating the trace id into our Splunk logs allows us to directly retrieve traces by entering the trace id into our Jaeger UI. This integration transforms isolated log entries into essential components of a broader control flow. By establishing this connection, we can gain valuable context regarding execution paths, enabling us to obtain insights into event data at specific moments. This streamlines the debugging process, as we can now understand the sequence of events leading up to an event, along with the state information.</p><p>Additionally, we enhanced integration between our logging and tracing tools by including log data as part of the Jaeger events. This change minimizes the need to context switch between the Jaeger UI and Splunk UI, as we can now capture any relevant information with the event in one cohesive view.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5RcGmp4Q3evekbKw2_U1pQ.png" /><figcaption><em>View of part of a Jaeger trace in Grafana. The logs section comes from logging statements in the code. Previously this information was only available in Splunk.</em></figcaption></figure><p>We also took advantage of Jaeger’s capability to attach exception tracebacks, allowing us to capture detailed stack information for any encountered exceptions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0k8iLoMFGfnTxQp7rSCthw.png" /><figcaption><em>In the Jaeger UI, we can now see the full stack trace for exceptions.</em></figcaption></figure><p>These enhancements provided several immediate benefits for any method annotated with tracing:</p><ol><li>When looking at logs in Splunk, we could easily access all duration metrics for the method.</li><li>We could easily visualize execution flow across domain boundaries including exception tracebacks.</li><li>Easy access to more logged state information made it easier to reproduce issues.</li></ol><p>With these improvements, we can now access the Jaeger trace corresponding to the sending of a specific SMS in under a minute. We can quickly ascertain the total execution time and confidently inform our internal or external customers if a delay originated in our systems or elsewhere. This efficiency contrasts sharply with the hours it previously took to answer the same question before enhancing our tooling!</p><h3>Customizing Sampling Strategies</h3><p>However, with a uniform random sample of only <em>0.01%</em> of requests, the likelihood of capturing the trace for any specific event was quite low. To address this limitation, we needed to enhance our sampling strategy. Certain events are inherently more critical, especially those involving high-volume customers or customers with a history of issues. Prioritizing these executions is essential for debugging potential regressions in our service level objectives (SLOs) and service level agreements (SLAs). This insight led us to focus on capturing more events from specific companies.</p><p>We decided to maintain the default sampling rate at <em>0.01%</em> while allowing for customizable sampling rates based on business rules. For instance, we could implement a <em>50%</em> sampling rate for company ABC.</p><p>To achieve this, we leveraged our internal hotsettings system, which enables us to dynamically configure settings stored in a Redis cluster and retrieved in real time as a hash set. This flexibility allowed us to define customized business rules through a JSON schema for any attribute we wished to trace, complete with its sampling rate. For example, if we wanted to log <em>50%</em> of all events for the customer with id ABC, our schema would look like this:</p><pre>{<br>    &quot;company_id&quot;: {<br>        &quot;ABC&quot;: 50<br>    }<br>}</pre><p>With this setup, we could dynamically adjust, in real time, how we capture events based on any specified attribute in our JSON schema, allowing us to increase or decrease the volume of information collected as needed.</p><h3>The Outcome: Efficient, Focused Observability</h3><p>The end result:</p><ul><li>Configurable sample rates let us limit excessive data capture while dynamically capturing the most relevant events.</li><li>Merging logs with traces provides a complete view of execution flow, covering start and end times, duration, input/output data, and cross-domain context propagation.</li><li>We can now quickly locate relevant Jaeger traces based on any attribute by first querying logs in Splunk.</li></ul><p>The combination of these benefits have helped us answer questions around specific latency from hours of looking at logs to a few seconds!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=262fbc51441d" width="1" height="1" alt=""><hr><p><a href="https://klaviyo.tech/understanding-timings-in-distributed-systems-262fbc51441d">Understanding Timings in Distributed Systems</a> was originally published in <a href="https://klaviyo.tech/">Klaviyo Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My experience releasing 3 failed SaaS products]]></title>
            <link>https://mmartinfahy.medium.com/my-experience-releasing-3-failed-saas-products-44e61cbde424?source=rss-66fc8e112439------2</link>
            <guid isPermaLink="false">https://medium.com/p/44e61cbde424</guid>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[bootstrapping]]></category>
            <category><![CDATA[failing]]></category>
            <category><![CDATA[startup-lessons]]></category>
            <dc:creator><![CDATA[Martin Fahy]]></dc:creator>
            <pubDate>Sun, 04 Apr 2021 01:29:27 GMT</pubDate>
            <atom:updated>2021-04-04T05:21:39.921Z</atom:updated>
            <content:encoded><![CDATA[<p>Build it and they will come! In the 1989 classic: Field of Dreams, an Iowa farmer has the vision to bulldoze a portion of his corn crop in order to create a baseball diamond in the middle of nowhere. To his credit, he does it spurred on by a loving spouse. After creating the field famous baseball players start appearing, but there’s a catch they’re all imaginary. Sound familiar? The movie is romantic and inspiring, that is till you launch a product that flops.</p><p>There’s a lot to be learned from the Field of Dreams from an entrepreneurial aspect, basically don’t do it. If you build it they probably won&#39;t come. Why would they, you built a baseball diamond in the middle of nowhere!</p><p>I got into this because I had just started reading about the time value of money, passive income, and entrepreneurship. I had realized early in my career that the real way to make the big bucks was to utilize a method that rewarded you non-linearly to the amount of effort you expended. It’s an appealing thought, laying on a beach as your product just prints money when you’re asleep.</p><p>From the successful founders I’ve researched, they’d probably be quick to tell you that nothing is really ever passive, if you create a business that can make 1k MMR it could probably make 3k MRR with more time and effort ad nauseam, that is till you’ve saturated the total addressable market. It’s going to be hard work every step of the way in fact Jason Cohen who’s released products with over 100 million MMR calls this the Technical Founder’s Dilemma basically your product gets so successful that you end up transitioning into a role you may not enjoy at later stages of growth. The point is passive is a lie it’s to sell you some course for $39.99. You could maybe hold off growing your company and milk the MRR but eventually, conversions will slow, competitors will come, and your product will slowly but surely fall short in SEO rankings.</p><p>What is failing? Is it releasing a product with 0 MRR, maybe 100 MRR, or a 1k MRR? Maybe, if you’re already a successful founder who created a company that sold for X millions and your new venture only makes you a couple thousand a year that’d probably be a flop for you, however, if you’re just starting out it could be an amazing success. Like all things failure is relative. To me failure is expecting an outcome, working hard to achieve that outcome, and falling short.</p><p>When I created my first product I just wanted to build something. I had ideas in the past and got lazy, none of them ever got shipped. I just wanted to take a side project from start to finish and release it, basically just to see what would happen. I created my app, built APIs, deployed it on AWS, and released it to the app store. To me, just shipping was a huge victory the work was done and now I wanted my check.</p><p>It was a hard but in retrospect grateful reality that exactly 0 people downloaded my app, not even my friends or family. When my family did try to download it, after much prodding, of course, the app crashed right on the splash screen. I tried to drum up support in niche subreddits only to have people tell me it was basically useless that they wanted a desktop version, not a mobile app. At the time I was heartbroken I had worked for months and the cumulation of all that work was $50 bucks a month in AWS bills and a shitty app that didn’t even launch. I called it quits shut the servers down and delisted the app retreating back to my den of solitude much as I came.</p><p>In the beginning, the goal was to ship and it felt great to actually release a product, intoxicating if I might. It hurt though that nobody found it even remotely useful. It was at this point that I learned the hard way that build and they will come is utter bull. It’s around this time I did finally start to do some mucking about the process itself but I hated the idea of marketing or doing sales especially as an introverted technical guy, I thought I was better than that. So a lot of what I read and listened to fell on deaf ears but I was picking up some of it.</p><p>I was restless. I didn’t care so much about creating a product that would actually make me money, I just wanted someone to use something I built. I couldn’t shake it I just wanted users, anything to know what that felt like. I had another quick eureka moment and ended up building a simple Chrome extension in a weekend, my second SaaS “product.” To my delight people used it, a lot of people used it. The downloads were small, to begin with, but with exactly 0 marketing in a matter of a month, I had a couple of hundred users. The level of excitement I felt was truly hard to describe even though only a few hundred people were using my very basic code I felt the deepest sense of satisfaction I ever had in my software career. I was obsessed with it reloading the stats to see how many people downloaded it. It was the first thing I did every morning.</p><p>Then came the first review the notification hit me right in bed as my groggy eyes scanned my phone. It was a summary of the review and it was bad. I forget the exact wording but basically, it was like “This app does not work as intended 0 stars.” It hit like a crash dummy flying into a brick wall at a crash site test. I was furious, despondent, and sad. I tried everything to get in contact with the user, I sent maybe 4 follow-up emails begging for me to help him or her fix their problem and to remove their review. Nothing, crickets. The downloads stopped happening, people stopped using it, and then churn started to occur. Again the highs led to a low and I removed the extension out of anger, knowing that basically it would never make a cent and it wasn’t worth my time updating it.</p><p>Again I had “failed” in the sense of not making a dollar of my product but I finally released something people wanted to use. I had read YC founders Paul Graham’s essays after my first flop and the one thing that stuck with me was startups are not about coming up with ideas and then building a product, <em>they’re about solving problems and building a solution for those problems. </em>That stuck with me this time around and I was actively solving a problem, my problem. Again with very little discovery about the market or users but at least it was a step in the right direction. However, as much as it feels good to have people actually use your product I realized it must feel even more amazing for people to <em>pay</em> for your product.</p><p>In the short span of a few months, I went from just ship something to understanding that I needed to build something useful. I needed to solve problems worth solving and needed to understand my customers, talk with them, and build something that would <strong>improve</strong> their lives. This was a hard lesson learned through blood and tears but there was no better way to learn. I had read that this is what I should be doing from the get-go long ago but didn’t heed much to that advice. Again I felt like I didn’t need all the “business” mumbo jumbo. Luckily I built my products myself because these would have been expensive lessons to learn if I had hired a development agency.</p><p>The latest and third product I think so far has been my greatest “failure” but equally the thing that has taught me more than anything. On the surface I did a lot of things “right” but in the wrong way, I’d so fatefully find out. After again going through the typical rinse repeat cycle of optimism to total despair with my last product; I spent a ton of time reading, listening, and watching content on how to run a bootstrapped SaaS.</p><p>I learned I needed to be talking to people from the get-go, understand them, and learn about their problems. Again this next idea was a problem I had, but before coding a lick I started doing customer validation or what I thought that was.</p><p>I was getting a list of users on niche forums and websites. I was talking to people explicitly about my idea gathering their input. Again great things to do, however, I was doing it the wrong way.</p><p>I was literally saying to people: “I’m looking to build an X app would you find that useful”. Hundreds and hundreds of responses to my posts were enthusiastic yeses I’d love to use that! I got bit by the build it bug and stopped right there in my tracks and started coding away, a huge mistake...</p><p>I had spent months working through a fully polished app that had many features. I was sending email updates to my list and happy to see people were engaging. Again I was at the peak of optimism, this was it! This is the app that I finally release to get paid users. The third time is the charm!</p><p>I slowly but surely started riling up my list by sending a sequence of launch emails, engagement was great, I was excited to launch.</p><p>Then I finally did it. I fired off the launch email to my list, notified the forums of my product, and almost instantaneously the users started flooding in.</p><p>It was exhilarating! In the first week, I had more users than I’d ever had in all my apps, people we’re using it first a hundred, then two, then three and then it stopped…</p><p>Then three hundred and ten, three hundred and twenty a few weeks later nothing. I looked at my metrics and while people were signing up almost nobody was actually using the product. Of course the whole time I was getting “great” feedback from my users: “The products great!”, “Great job keeps working at it”, “I really like the design.” Nobody though was actually using it? Why was nobody using it? The people who did use it may be used it for a max of ten minutes before never using it again. Why the heck was nobody using it?</p><p>Soon I started getting feedback on forums along the lines of “The product looks interesting but I don’t really want to sign up for it without seeing the product first.” Again this was a free product at the time and all the user had to do was click the Google or Facebook OAuth button to log in, or enter their email but they refused to do so. <strong><em>They didn’t trust me</em></strong>. I started to look deeper at my metrics and realized I had a huge bounce rate maybe over 3k or so visitors at the peak and a bounce rate of 80%. I had sustained over 1k users hitting my landing page a day for the first week and only 300 users actually even signed up to use the app and when they did they almost immediately churned. Nobody trusted me enough to actually try my app to begin with or maybe it wasn’t actually a painful enough problem as I initially thought.</p><p>Of course, my first thought as a technical founder was that there weren’t enough features so I furiously started creating as many additional features as I could. I fired up my email list and told them all about the new features and why they were so awesome. The engagement was bad and it was the first time people started actively unsubscribing from the email list, over 10% in my first email after launch, with more coming on each subsequent email.</p><p>After getting minimal traffic for a period of a week or two after releasing the new features I realized I needed to market my app to get the traffic up. I didn’t even connect the dots on my metrics perhaps it was denial, but I thought if I can just throw more proverbial shit at the fan I’d get more users, which was technically true. I started to “market” horrendously going to more sub-niches again getting a quick little spike of traffic and users to high churn and bounce. I created a Youtube channel and blog to increase SEO and started to actually get involved with learning how to market realizing its importance something I was so happy to dismiss earlier in my entrepreneurial journey much to my chagrin. My ads were useless and the bounce rate only increased after doing outbound marketing, if you could even call it that.</p><p>It was then when it happened. To my credit, at this time I would learn and do anything to get more users to my site, to the point where I was getting banned from multiple forums for posting after people told me to get lost for promoting content. I had fully embraced marketing and was furiously reading content and trying to increase my SEO ranking for long-tail keywords in the hopes of it driving traffic. It was then when I learned about backlinks and why they’re important for ranking. I put my domain in and to my surprise I had multiple backlinks quite a few in fact. Initially, I was stoked! Then my heart almost left my chest, the backlinks looked sketchy and the title of some of the backlinks was offensive. I clicked on the link and it led me to a forum I had not posted on were over twenty or so people were just tearing both me personally and my app apart. I will refrain from summarizing but basically, it was extremely hurtful. I cried.</p><p>It’s then why it clicked why the bounce rate was high and my conversion rate was low people just didn’t want to use my app, they actively hated it to the point of making blog posts about why it was such an awful idea. I felt crushed my cycle of optimism to despair was at the lowest despair I had ever felt. I was working on a blog post the night I found the backlink as I choked back tears I kept typing telling myself they were just haters and to not give up. I couldn’t keep typing though and I stopped working on it from that day onwards.</p><p>I was hurt for a week or so, I kept the site live to see if any of my metrics would improve, a week turned into two. The metrics got worse and worse, and a few weeks later there wasn’t a sign-up in over three weeks. It’s then when I killed the product destroying all my users saved progress for anyone who had actually used the product. It was relieving since the serverless bills had racked up too close to $200 bucks. <em>Let me just say there is no worse fate for a creator than their product to silently be scrapped without even the most minor of eeks or anger from its users</em>. Nobody cared it was gone, people just barely cared when it was there, but people especially did not give a fuck when it was gone.</p><p>So why did it “fail”? We’ll it failed because while I got validation of my idea I was doing it the wrong way.</p><p>Telling people: “Hey this is my app idea would you use it?” Is a clear violation of the Mom Test created by Rob Fitzpatrick a great book I’d highly recommend. In short just because users or potential customers tell you your product is something they’d use you don’t really know that’s the case without some action from that potential customer, whether it&#39;s signing up for a beta or better yet trying to pay for the product before its built.</p><p>I had never really understood building a sales funnel before launching a product but after my last failure, I totally get it. You want people to try and pay and use your product before you even start building it, that way at the very least when you release it you have paid customers and you can also validate it’s a true problem. A problem people are so frustrated with that they’d be willing to potentially pay you before it even exists.</p><p>People initially told me they liked the idea because I was exposing my ego and it was easier to tell me they’d use it than telling me to buzz off. They also were looking to solve a broad goal which my product could have helped them with but the execution of which lacked to really compel the user to truly use it. This explained why people would use the product for all of two to three minutes then churn.</p><p>Another point of not building the sales funnel upfront is when I released my product to people who were highly interested in it, my top demographic wouldn’t even sign up for it because we had not built trust. These people were in the middle of the funnel and I thought they’d just automatically close i.e. sign up once the product was out there. Some did but the majority didn’t.<em> If I had marketed the blog upfront, then built my funnel, and sold them later even if the product wasn’t built I would have had a much higher chance of realizing first if the product was truly worth building. I’d have real data as to whether people would buy it or not.</em></p><p>Also by marketing upfront your building SEO months before release so you can hopefully piggyback on some of the small gains you’ve made there to keep the top of the funnel supplied once the big initial spike occurs from your email list. It’s easy to get a huge spike in engagement at launch if you have a list it’s very hard to keep that level of engagement going after.</p><p>Some customer validation is good but to get real results you need serious effort in understanding your customer&#39;s problem and empathizing with them. You need to understand your customer intimately you should exactly know the type of person you’re building this for and what they&#39;re trying to accomplish. You should be providing value to them every step of the way through your funnel. This builds trust and a dialog between you and the customer, they begin to respect you and your offering and think you can help them. Then you actually do. Your product doesn’t need to be polished if it solves the customer&#39;s problem accurately. This is I think is the main point lost in the “ship it” culture. An MVP doesn’t need to be pretty but it absolutely must, without doubt, solve your customer&#39;s problem and it should not be shipped any time sooner or later than when it can accomplish that.</p><p>If you actually want to do this spend as much time reading every resource on sales and marketing you can. Hubspot is a great resource to learn from. If you’re a technical founder let me just say code is evil and you wanting to write it is bad. This is not a side project if you just want to code something go do it but don’t expect it to sell. If you want to actually create a business you need to realize coding is both the easiest of all the things that need to be done and also the least important. Let me say this again you can run a successful SaaS company with the simplest of Python scripts if you can market and sell that solution. Marketing and selling is the crusp of what it means to be a bootstrapper, think of yourself foremost as a marketer who knows how to code rather than a coder who knows how to market. You should be spending as little time coding and as much time marketing, building leads, and optimizing your funnel.</p><p>Lastly, features are useless if you’ve got a churn problem features won&#39;t solve it. If you got a bounce problem features won&#39;t solve it. If you have a traffic problem features won&#39;t solve it unless you’re improving SEO. Remember code is evil.</p><p>So what&#39;s next? We all see everyone’s highlights real on social media and it&#39;s hard not to feel bad about failure, but what do you really lose as a bootstrapper when your product flops: maybe some time and ego. I’ve learned so much so far on this journey and happy to be reminded that others have failed many more times before finally getting their “overnight success.” We all see the result but rarely do we see the messy side of how people got where they are. I’m spending a lot of time decompressing and focusing on whatever the next thing will be. Reading as many resources as I can and searching for my next opportunity. If my journey has so far shown you anything it&#39;s progression. A year ago I scoffed at marketing and sales I thought it was sleazy. Today I have the utmost respect for those professions and doing my best to learn and get good at them myself. Bootstrapping is a journey of self-actualization it&#39;s so far been the hardest thing I’ve done in my life but easily the most rewarding and I haven’t even made a dime from it yet! I’m excited to build my next product, only of course after trying to sell it first.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=44e61cbde424" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>