<?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 George Sequeira on Medium]]></title>
        <description><![CDATA[Stories by George Sequeira on Medium]]></description>
        <link>https://medium.com/@sequeira?source=rss-268cc458acc7------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*h3jzUlycB9nPtv6a-fz_eQ.png</url>
            <title>Stories by George Sequeira on Medium</title>
            <link>https://medium.com/@sequeira?source=rss-268cc458acc7------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 22:08:21 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@sequeira/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[Zapier: The $5B Unbundling Opportunity]]></title>
            <link>https://medium.com/@sequeira/zapier-the-5b-unbundling-opportunity-ee28dccca4b6?source=rss-268cc458acc7------2</link>
            <guid isPermaLink="false">https://medium.com/p/ee28dccca4b6</guid>
            <category><![CDATA[api]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[software]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[business]]></category>
            <dc:creator><![CDATA[George Sequeira]]></dc:creator>
            <pubDate>Mon, 11 Apr 2022 20:04:34 GMT</pubDate>
            <atom:updated>2022-04-12T18:26:54.292Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*HFzE3FxYj-HA_8ZK.png" /><figcaption>(Credit: <a href="https://zapier.com/learn/zapier-quick-start-guide/quick-start-create-zap/">Zapier</a>)</figcaption></figure><p><em>Note: </em><a href="https://news.ycombinator.com/item?id=31003071"><em>This post got picked up on hackers news </em></a><em>— a bunch of great conversations were had around the topic.</em></p><p>In September 2019, A16Z <a href="https://a16z.com/2019/09/11/platforms-verticals-unbundling/">described</a> a strategy for identifying new startup opportunities: look at broad horizontal platforms that are near their breaking point. Horizontal platforms usually succeed in providing simple interactions that support a variety of verticals, think Craigslist (renting apartments, finding jobs, creating events), eBay (selling tickets, cars, shoes), and Reddit (chat for communities around anything).</p><p>As these platforms scale, niche use-cases grow big enough in usage and value to create opportunities for businesses to deliver more tailored products. As the number of people looking for jobs on Craigslist grew, companies like Indeed and ZipRecruiter saw an opportunity to create a better experience. The process of companies building products that strip away use-cases from horizontal platforms is called “unbundling”.</p><blockquote>The process of companies building products that strip away use-cases from horizontal platforms is called “unbundling”.</blockquote><p>In their post, A16Z lists Craiglist as an example of a platform that has been unbundled and points towards LinkedIn as one currently undergoing the process (<a href="https://a16z.com/2019/09/11/platforms-verticals-unbundling/">2019</a>). Given the explosive growth of SaaS products and the growing need to have them inter-operate, the next big platform to be unbundled is Zapier.</p><h3>What is Zapier?</h3><p><a href="http://zapier.com/">Zapier</a>, valued at <a href="https://www.forbes.com/sites/alexkonrad/2021/03/08/zapier-bootstraps-to-5-billion-valuation/?sh=111366b81c8e">$5B</a>, is an integration platform that gives non-technical end-users the ability to connect SaaS products that aren’t natively integrated. With Zapier’s point-and-click solution, a user can create business processes by connecting SaaS products in just a few clicks.</p><p>For example(s): Need to have your website sign-ups send an email? Send an email from Mailchimp on every TypeFrom submission. Need to create todos every time a new deal comes into your Salesforce? Create a ticket in Jira when a new deal lands in Salesforce. Want to know when your delivery is on its way? Send a Slack message when an email has the term “Your order is on its way” in the subject line. All in Zapier. All without code.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*4YKmlTzn19sz4-K3" /></figure><p>At the time of this writing, there are over 3,000 SaaS integrations you can use on Zapier. Whatever SaaS products you’re using, they probably all connect with Zapier. From small e-commerce stores automating marketing emails by connecting MailChimp and Stripe to startups tracking their sales teams by plugging Salesforce into Asana, Zapier has grown to over <a href="https://www.drift.com/blog/how-zapier-grew/">3M</a> active users and generates over <a href="https://sacra.com/research/zapier-netflix-productivity/#:~:text=Zapier%20hit%20%24120M%20in,50%25%20from%20the%20year%20prior.">$125 million</a> in annual revenue.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*tU2ipErk3HXIY4MT" /><figcaption>(Source: <a href="https://sacra.com/research/zapier-netflix-productivity/#:~:text=Zapier%20hit%20%24120M%20in,50%25%20from%20the%20year%20prior.">Sacra</a>)</figcaption></figure><h3>Zapier’s Advantages</h3><p>Zapier’s secret is its simple yet powerful paradigm of “triggers” (something happens in one product) and “actions” (do something in another product) which is controlled all by mouse-click. You don’t have to think about authentication, infrastructure, REST APIs, queries, or parsing; you can translate your needs into a one-for-one step process in Zapier. For example: “I wish when someone signed up on my site, I could send them an email”, is a quick 4 step process.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*c_WnPr34INKknMwn" /></figure><p>Zapier’s advantage as a platform is (was) their first-mover strategy. When I first heard of Zapier a decade ago, they had over 100 APIs and I thought that was more than I’d ever need. Over time, as more SaaS companies came online, Zapier continued to grow to <a href="https://zapier.com/platform">over 3,000</a> APIs! As businesses continue to rely on SaaS for their operations, Zapier is the go-to place to make SaaS products talk to each other.</p><p>Zapier’s growth strategy has been pretty simple: build more integrations and fulfill more use-cases; fulfill more use-cases, and attract more customers. While Zapier has slowly created 3,000+ API integrations over 10 years, the number of SaaS products that an organization uses has gone from <a href="http://pages.bettercloud.com/rs/719-KZY-706/images/2020_StateofSaaSOpsReport.pdf?mkt_tok=NzE5LUtaWS03MDYAAAF8LQdmoC7u54xbqxNwp0au4Zk7SiYaaqq2vupXFxCvaP5vY8gSQtlGFsUsRI8oj5Fl2m5PwIZUUAlzVZL_-hUEQ2RdNqgEzDAmZA5bZtowS_v-zMs">8 in 2015 to over 80 in 2020</a>.</p><blockquote>Zapier’s growth strategy is continuing building more integrations on top of their point-and-click platform.</blockquote><p>This isn’t just a concern of monster companies like Coca-Cola or Deloitte anymore, companies smaller than 50 employees <a href="https://explodingtopics.com/blog/saas-statistics">leverage 16 SaaS products</a> on average in their organizations.</p><h3>Breaking Point</h3><p>With the average number of SaaS tools used in a business increasing 10x from 2015 to 2020, Zapier’s breadth and simplicity are reaching a breaking point.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*A-TwwljWetz-OLrT" /><figcaption>(# of SaaS Apps used on average by organization. Source: <a href="http://pages.bettercloud.com/rs/719-KZY-706/images/2020_StateofSaaSOpsReport.pdf?mkt_tok=NzE5LUtaWS03MDYAAAF8LQdmoC7u54xbqxNwp0au4Zk7SiYaaqq2vupXFxCvaP5vY8gSQtlGFsUsRI8oj5Fl2m5PwIZUUAlzVZL_-hUEQ2RdNqgEzDAmZA5bZtowS_v-zMs">BetterCloud</a>)</figcaption></figure><h3>Breadth</h3><p>While Zapier can connect with virtually every API, it doesn’t support any single API extremely well. Something as simple as being able to update a Shopify order requires <a href="https://community.zapier.com/featured-articles-65/how-to-update-an-order-in-shopify-6696">a 500-word tutorial </a>to get working. Even simpler: doing individual “actions” over a list of items from a “trigger” (think sending an email to each of the customers you get from a purchase report) wasn’t a feature <a href="https://zapier.com/help/create/other-functions/loop-your-zap-actions">until 2021</a>. There are over <a href="https://community.zapier.com/search?q=workarounds">880 posts</a> on Zapier’s forums detailing how to get API-specific use-cases to work. Zapier’s breadth is resulting in complicated workarounds to get the most out of APIs.</p><blockquote>Over 880 posts on Zapier’s forums detail how to get API-specific use-cases to work.</blockquote><h3>Simplicity</h3><p>Back when business owners had simple aspirations for their digital products (e.g. add a person’s details into a Google Sheet after they buy something through Stripe), a simple point-and-click would suffice. With the average company leveraging over 80 SaaS products, modern organizations require more complex integrations that can become unwieldy with the primitives provided by Zapier. It’s becoming more common for the management of Zapier workflows to be <a href="https://www.upwork.com/ab/profiles/search/?q=zapier&amp;user_pref=2">outsourced</a> or managed by a technical team.</p><blockquote>Making 80 SaaS products work together requires the help of outsourced talent or an in-house technical team.</blockquote><h3>Unbundling Zapier</h3><p>Just like Craigslist, eBay, and Reddit before it, Zapier is in the perfect position to be unbundled. As a large horizontal platform with a growing number of users, there are clusters of Zapier users in need of focused alternatives. People have noticed this opportunity already:</p><ul><li><a href="http://workato.com/">Workato</a> and <a href="http://make.com/">Integromat</a> are working on “Zapier for Business” with features like SOC2-compliance, role-based-authentication, and on-premises support.</li><li><a href="https://runalloy.com/">Alloy</a> and <a href="http://anvyl.com/">Anvyl</a> are focusing on vertically specific integration platforms (E-commerce and Supply-Chain respectively) that support just 100s of SaaS products but make common use-cases simpler for their respective industries.</li><li><a href="https://airtable.com/">Airtable</a> and <a href="https://reclaim.ai/">Reclaim</a> have replaced many integration use-cases by creating feature-rich versions of the products most commonly involved in Zapier integrations (spreadsheets and calendars respectively).</li><li>Finally, <a href="https://pipedream.com/">Pipedream</a> focuses on better support for complex Zapier use-cases by providing a platform that software engineers can use to create more technical and nuanced integrations.</li></ul><p>These are just a few ways that Zapier is being unbundled but there’s plenty of unbundling left. I’m personally building a vertical-specific product that reduces reliance on integration platforms altogether. More to come soon ;)</p><p>If you’re interested in collaborating or chatting about the journey, please reach out! Good luck friends!</p><p>More writing at: <a href="http://georgesequeira.com/">georgesequeira.com</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ee28dccca4b6" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Great teams hire “High-Clarity + High-Energy” people]]></title>
            <link>https://medium.com/@sequeira/great-teams-hire-high-clarity-high-energy-people-84d34dc1d90a?source=rss-268cc458acc7------2</link>
            <guid isPermaLink="false">https://medium.com/p/84d34dc1d90a</guid>
            <category><![CDATA[lessons-learned]]></category>
            <category><![CDATA[people]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[hiring]]></category>
            <dc:creator><![CDATA[George Sequeira]]></dc:creator>
            <pubDate>Fri, 21 Jan 2022 23:27:29 GMT</pubDate>
            <atom:updated>2022-01-21T23:27:29.870Z</atom:updated>
            <content:encoded><![CDATA[<p>To build a successful startup, your number one priority is to hire great people. Being energized by the people you work with translates to everyone doing their best work. When I was building my company in 2020, I wanted our edge to be our employees.</p><p>After examining my years of building <a href="https://alexa.amazon.com/spa/index.html">world changing products</a>, <a href="https://www.bluecore.com/">scaling a unicorn</a>, and <a href="https://linkedin.com/in/georgesequeira">working at various startups</a>, I found that in my career, periods of amazing outcomes correlated with having high-clarity + high-energy colleagues. What’s more, when organizations had low-clarity + low-energy people, my work was slower and emotionally draining. In order to build a world-class team, I needed to find a repeatable measure of high-clarity + high-energy people; this led me to develop the Clarity + Energy Matrix.</p><blockquote><strong><em>Amazing outcomes correlate with having high-clarity + high-energy people.</em></strong></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ng_wyyiWwq29UXptS7Q7mw.png" /></figure><h3>What is clarity?</h3><p>Clarity is how effective someone conveys information through their words and actions. High-clarity people create understanding in everything they say and do, they require direction but don’t require perfect information to act. Low-clarity people create confusion and distraction, usually they ask (or are told) multiple times how to correctly satisfy requirements.</p><p>In startups, time is the most valuable asset. Time spent building the wrong thing is worse than not building anything at all. High-clarity people reduce the risk in building the wrong thing or building for the wrong reasons. This leads to more autonomy and faster innovation speeds.</p><blockquote><strong><em>High-clarity people promote more autonomy and higher innovation speeds.</em></strong></blockquote><h3>How to measure clarity?</h3><p>Judging someone’s clarity is not a matter of changing your process as much as it is changing what you perceive in your existing interviews. I added the following steps to our culture questions to judge clarity</p><ul><li>Through the question, take notes on what you perceive the motivation, the key take-aways, the goals, the result, of each situation.</li><li>At the end of the question take a couple minutes to review your understanding with the candidate. This is separate from the actual contents of the question and should be evaluated independently.</li><li>Take the opportunity to have the candidate correct anything you may have missed.</li></ul><p>If at the end, you felt that you continually kept learning without backtracking, that your interpretation was easy to write down, and that the candidate actively sought to help you understand, you’ve got yourself a high-clarity person.</p><h3>What is energy?</h3><p>Energy is the effect that a person gives others. This energy can be curiosity, inspiration, creativity, and yes even excitement. Different than the “beer test”, you aren’t looking for someone to entertain you, this is about bringing someone onboard that makes the whole team better.</p><p>Startups are a scarce resource environment. They require being able to create something out of nothing with fewer resources than you’d like. The only way to achieve the impossible is to have a team of people who’s whole is greater than a sum of their individual parts. High-energy people promote a growth-mindset and increase the breadth of problems your company can solve.</p><blockquote><strong><em>High-energy people promote a growth-mindset and increase the breadth of problems your company can solve.</em></strong></blockquote><h3>How to measure energy?</h3><p>The question you have to answer with energy is: will this person push the team positively in some way?</p><p>Not everyone has the same superpower, a few examples I’ve seen of extraordinary energy:</p><ul><li>The curious designer that influenced others to think differently about the right answer to our customer’s problems.</li><li>The resilient leader who created outcomes from ambiguous situations that others thought were impossible.</li><li>The detail-oriented engineer who kept us honest and helped create one of the smoothest products I’ve built.</li></ul><p>As this is an interpersonal trait, it requires getting feedback from multiple people. As your team members evaluate their interview slots, at the end, they should answer: what will this person push me in?</p><blockquote><strong><em>What will this person push me in?</em></strong></blockquote><p>Review all your answers at the end. If most (&gt;%50%) of your team feels like they will be pushed in some way: you’ve got a high-energy person.</p><p>As you continue to hire, you’ll notice that your team’s standards will continually get higher as you keep bringing in people that make everyone better. This flywheel creates a culture of collaboration, growth-mindset and achieving the impossible. That sounds like a world-class team :)</p><blockquote><strong><em>High-energy people create a cultural flywheel of collaboration, growth-mindset and achieving the impossible.</em></strong></blockquote><h3>The Clarity-Energy Matrix</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ng_wyyiWwq29UXptS7Q7mw.png" /></figure><p>So by now you should be convinced that you need high-clarity + high-energy people! When I developed this matrix, I loved that it easily slid in to our interview processes but vastly increased our understanding of candidates. At the end of each candidates interview loop, I would write and share feedback with the team and then simply put the candidate in the matrix. The top-right people were consistently our 10x’ers and were a “yes” well after their start date.</p><p>This matrix can also give you insight into your current company environment:</p><ul><li>As an employee, if your leaders are not in the top right, you should leave.</li><li>If you’re a manager, if your direct reports aren’t in the top right, you have some work to do.</li><li>As an executive leader, if you don’t feel your other leaders are in the top-right, it’s time to get them some coaching.</li></ul><h3>What about low-clarity or low-energy?</h3><p>I get it. Sourcing good candidates is hard. What do you do when you have a competent candidate that lacks high-clarity or high-energy? Or even if they have mid-clarity and mid-energy?</p><p><strong>Nothing.</strong></p><p>In my experience these folks end up lacking self-motivation and self-discipline, resulting in dulling the edge on your would-be world-class team. One wrong hire can kill your team; anything but high-energy + high-clarity trends towards mediocrity. Zero high-growth startups have mediocre teams, you shouldn’t.</p><blockquote><strong><em>Zero high-growth startups have mediocre teams, you shouldn’t.</em></strong></blockquote><h3>Conclusion</h3><p>90% of startups fail. There are a bunch of reasons that basically boil down to one thing: the right people weren’t in the room.</p><p>In order to build a world-class team you need high-clarity + high-energy people. Use the matrix and hire everyone in the top right. Good luck friends!</p><p>More writing at: <a href="http://georgesequeira.com/">georgesequeira.com</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=84d34dc1d90a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Breaking into B2B SaaS]]></title>
            <link>https://medium.com/@sequeira/breaking-into-b2b-saas-35352030ba85?source=rss-268cc458acc7------2</link>
            <guid isPermaLink="false">https://medium.com/p/35352030ba85</guid>
            <category><![CDATA[startup-lessons]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[saas]]></category>
            <category><![CDATA[software]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[George Sequeira]]></dc:creator>
            <pubDate>Thu, 18 Nov 2021 15:39:21 GMT</pubDate>
            <atom:updated>2021-11-18T15:39:21.255Z</atom:updated>
            <content:encoded><![CDATA[<p>Building a company is hard! Unless you have conviction in an area through prior experience, figuring out your next big idea is a daunting task. Do you build an app that finds the best parties? Everyone likes to party, right? Wrong. So how do you get started searching for ideas?</p><p>At the start of 2020, I had zero clue on what I wanted to solve… I spent 2 months working on a podcast app that ended up with a total of 10 users and I had 8 months left in my runway. I knew that I wanted to solve a big problem but following the advice of “<a href="https://www.ycombinator.com/library/6e-how-to-evaluate-startup-ideas">solve your own problems</a>” always led me to consumer fitness or health apps, which is a heavily saturated market. After months of building consumer solutions (without problems) and trying to figure out how to get paid for them, I realized that I needed to focus on solving problems for people who would pay for solutions: businesses.</p><p>Focusing on businesses allowed me to expand my scope from “my own problems” to “problems I’ve encountered or known about from work” that could be applied across industries. After making the switch and focusing on business problems, I was able to research 6 industries/markets and end up founding a company in the automotive industry within 4 months. I found that if you want to discover many ideas fast, there is one business model that directly correlates with solving problems for people: B2B SaaS.</p><h3>B2B SaaS</h3><p>B2B SaaS stands for business-to-business software-as-a-service. Think Slack. As a B2B SaaS company you are selling cloud-based software to other companies. It could be super-specific like email-solutions for e-commerce companies like <a href="http://bluecore.com/">Bluecore</a> or something as broad as managing work tasks like <a href="http://monday.com/">Monday</a>. Businesses need software solutions and it doesn’t show signs of slowing down (spend on SaaS is up <a href="https://www.blissfully.com/blog/saas-statistics/">50% from just two years ago</a>).</p><p>You may wonder what’s left to automate but it’s important to note that <a href="https://www.bmc.com/blogs/saas-growth-trends/">22% of small businesses</a> haven’t invested in any SaaS at all. It’s not an if but when they will take advantage of SaaS solutions. And while <a href="https://future.a16z.com/software-is-eating-the-world/">the technology necessary to transform all industries</a> has been available for over a decade; risk-averse, low-innovation, and human-centric industries are just realizing that they can leverage SaaS to reduce costs and optimize workflows.</p><h3>Finding the industries and markets</h3><p>Since B2B SaaS focuses on selling to businesses, the industries to focus on are those that make a lot of money and/or are growing. If you’re already interested in an industry, be sure that it is one of these two. Use tools like <a href="https://www.cbinsights.com/">CBInsights</a> to analyze overarching trends in industries and identify viable options. Also, you can look at industries that VCs (eg. <a href="https://www.sequoiacap.com/companies/">Sequoia</a>) are investing in or are requesting for startups in (e.g. <a href="https://www.ycombinator.com/rfs/">YCombinator RFS</a>). It’s their job to bet on the right industries, so leverage some of their homework for your own benefit.</p><p>Once you have your top-level industries of interest, dive into industry specific reports (e.g. google search for ‘Carbon Removal trends 2021’), ping industry experts you find on blogs, and start pinging folks in your network (e.g. <a href="http://linkedin.com/">LinkedIn</a>, <a href="http://beondeck.com/">OnDeck</a>) to understand where the opportunities are. Again, you are looking for big industries with some parts that are either due for a change or are already experiencing some growth.</p><p>Let’s try this for one industry!</p><p>I looked over YCombinator’s <a href="https://www.ycombinator.com/rfs/">Request-For-Startup</a> page and found that YC was focusing investments in <a href="https://www.ycombinator.com/rfs/#ai">AI</a>. One of their listed examples for AI disruption is manufacturing. I quickly validated that not only is the market huge at <a href="https://www.statista.com/outlook/io/manufacturing/united-states">$7.7T</a> but in 2022 will <a href="https://www.statista.com/outlook/io/manufacturing/united-states#revenue">be 2% larger</a> than pre-pandemic times. I googled and found Deloitte’s “<a href="https://www2.deloitte.com/content/dam/Deloitte/us/Documents/energy-resources/us-2021-manufacturing-industry-outlook.pdf">2022 Manufacturing Outlook Report</a>” which reports that over 40% of manufacturing executives are investing in “operational efficiency solutions such as: automating processes, increasing machine insights with IOT devices (think small sensors)…. sensors to track assets (<a href="https://www.thetileapp.com/en-us/">Tile</a> for a million dollar drill), and real-time insights on tracking maintenance of machines.” At the end of the report there are some sources and folks listed; I would send linkedin requests and emails to these people to learn more.</p><p>Just from the cursory research, I would guess that there are analytics and data management opportunities for manufacturing businesses with a long-term play at taking the data gathered to automate some of their manual processes (AI!).</p><p>Fun! One general opportunity found!</p><p>….. btw after reaching out to someone in my network regarding manufacturing, I was pointed to the expected growth of additive manufacturing <a href="https://www.metal-am.com/am-market-forecast-to-reach-51-billion-by-2030/">15% year-over-year</a> from now through 2030 ;)</p><h3>Finding an entry point into the industry</h3><p>So, let’s say you’re hooked now! You are all into providing SaaS for manufacturing… How do you break in?</p><p>Talk to people.</p><p>Don’t fall into the trap of being a laptop/armchair genius. You can do all the research on the internet you want, but at the end of the day, your business is going to sell to people and they are going to need their problems fixed. So start emailing and calling people to learn about the industry.</p><p>If you are already exposed to the industry via past experiences, it’s time to start calling these folks and <a href="https://www.amazon.com/Startup-Playbook-Founder-Founder-Techstars/dp/1119708516/ref=asc_df_1119708516/?tag=hyprod-20&amp;linkCode=df0&amp;hvadid=459726176530&amp;hvpos=&amp;hvnetw=g&amp;hvrand=7678528804802691027&amp;hvpone=&amp;hvptwo=&amp;hvqmt=&amp;hvdev=c&amp;hvdvcmdl=&amp;hvlocint=&amp;hvlocphy=9067609&amp;hvtargid=pla-918953553248&amp;psc=1">running discovery interviews</a>.</p><p>If you are coming to an industry with zero exposure, don’t worry. Someone you know or someone that knows someone you know will have worked in the industry. Also, look at startups that are solving problems in the industry and ping founders/employees via LinkedIn, see if you have friends that work at those companies. Start reaching out to all of these people and be <a href="https://www.georgesequeira.com/writing/go-from-nothing-to-funding-in-9-months-learn-out-loud">transparent about your learning process</a>; people will help you if you give them a chance.</p><p>Your main goal is creating a network of industry insiders for your company.</p><p>One cheat-code that exponentially increases your network is to ask the following question from anyone you meet:</p><p>“Who are 3 people you think I should talk to that are either working in this industry or are innovating in this industry?”</p><p>Boom. 1 goes to 3, 3 goes to 9, 9 goes to 27. You are bound to find some information and problems there.</p><h3>Understanding the industry</h3><p>I know, it’s wild! Just 2 weeks ago you thought manufacturing was a cool space to investigate and now you have 18 people on the calendar to talk to. You may be thinking: “I don’t know anything about manufacturing, how am I going to solve their problems?”</p><p>No worries. All you have to do is listen. People love talking about their jobs, especially if you are promising a better path forward. Your goal is to learn as much as possible about:</p><ul><li>The industry.</li><li>The specific market the people work in.</li><li>The key decision makers in the market.</li><li>The problems this person faces.</li><li>The problems the market is facing.</li><li>Who are other people they can point you to to learn more about the above.</li></ul><p>Your initial conversation won’t be enough. This is the start of a beautiful and new friendship. Don’t forget to take them out for lunch, coffee, or send them a thank you gift (or whatever you do for friends).</p><p>I recommend reading <a href="https://www.amazon.com/dp/B01H4G2J1U/ref=dp-kindle-redirect?_encoding=UTF8&amp;btkr=1">The Mom Test</a> to help structure your conversations. At the end of these chats, keep the door open with “This was super helpful, I’d love to keep chatting if you’re up for it!” and always look for ways that you may be able to help them (outside of solving their problems of course!).</p><p>You’re on your road to being an expert!</p><h3>Conclusion</h3><p>B2B SaaS is a great business model because it’s based on the fact that someone at work could use software to make their job easier and/or more effective. By doing top level research you can figure out what industries and markets are likely to have costly problems but in the end, building a company is about talking to people! Your job is to find the right people to talk to; the sooner you understand who the major players are, who key decision-makers are, and what problems exist for you to solve, the closer you’ll be to building a high-growth company.</p><p>If you ever need a sounding board, feel free to reach out! ​​Good luck friends!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=35352030ba85" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Finding startup ideas that anticipate the future]]></title>
            <link>https://medium.com/@sequeira/finding-startup-ideas-that-anticipate-the-future-a81086ad1680?source=rss-268cc458acc7------2</link>
            <guid isPermaLink="false">https://medium.com/p/a81086ad1680</guid>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[business]]></category>
            <dc:creator><![CDATA[George Sequeira]]></dc:creator>
            <pubDate>Mon, 25 Oct 2021 16:29:53 GMT</pubDate>
            <atom:updated>2021-10-25T16:29:53.788Z</atom:updated>
            <content:encoded><![CDATA[<p>After years of building, advising, and investing in startups, I’ve realized that while I love supporting high-growth companies, I am most passionate about companies that solve the problems of the future. For example, it’s incredible that while the Biden administration <a href="https://www.whitehouse.gov/briefing-room/statements-releases/2021/08/05/fact-sheet-president-biden-announces-steps-to-drive-american-leadership-forward-on-clean-cars-and-trucks/">recently passed more incentives</a> to accelerate electric vehicle adoption, companies like <a href="https://chargepoint.com/">ChargePoint</a> and <a href="https://www.weavegrid.com/">WeaveGrid</a> have been working for years to support the incoming energy demands.</p><p>It’s obvious in retrospect but how did these companies guess the future? How can they tell what the solutions of tomorrow look like? Coming up with good startup ideas is hard, add on top of that trying to guess the future, and it becomes an insurmountable task. After speaking with founders at these types of companies, I’ve learned that if you are looking to create a future defining company, you must find trends as they start, you’ve got to look for inflection events.</p><h3>What is an inflection event?</h3><p>An inflection event is a major turning point that enables a dramatic change and opens up the world to new possibilities. It signals a tectonic shift within industries, companies, economies and more. The event typically creates new opportunities for innovation rather than direct outcomes.</p><h3>What are examples of inflection events?</h3><p>Inflection events can include technological advancements, regulatory changes, or societal changes. I’ve put together pivotal events across each of these industries and businesses that were a result.</p><h4>Technological Advancements</h4><p>Whether it’s the <a href="https://www.apple.com/newsroom/2008/07/14iPhone-App-Store-Downloads-Top-10-Million-in-First-Weekend/">App Store </a>enabling <a href="https://www.uber.com/">Uber</a> or the creation of <a href="https://ethereum.org/en/">Ethereum</a> enabling <a href="https://opensea.io/">NFTs</a>, from logistics to art, technological advancements literally unlock solutions that were previously impossible. One recent example is the launch of the generative text model <a href="https://openai.com/blog/gpt-3-apps/">GPT-3</a>: grossly simplified by example, it allows blog posts to be generated from 2 lines of texts, software to be written by translation from english text, and even create an image from a provided description. The future unlocked by GPT-3 is just beginning. One such company is <a href="https://compose.ai/">Compose.AI</a> which reduces the amount of effort you take to write emails. You can imagine that it will eventually go from emails, to documents, perhaps even to reduce and short-cut slack conversations!</p><h4>Regulatory Changes</h4><p>The term “regulation” doesn’t usually invoke thoughts of future-building, but since the government’s job is to create a better future for its citizens, sometimes laws are passed to incentivize the private sector to help build that future. When multiple states <a href="https://www.healthaffairs.org/do/10.1377/hpb20160815.244795/full/">passed laws</a> in 2016 that forced insurance companies to cover telehealth costs, startups like <a href="https://www.forhims.com/">Him</a>+<a href="https://www.forhers.com/">Hers</a> and <a href="https://www.talkspace.com/">Talkspace</a> were able to target specific health-issues in a more convenient manner at reasonable prices.</p><p>Even more recent is <a href="https://www.marketplace.org/2021/04/05/new-rule-gives-patients-access-to-all-of-their-medical-records-and-notes/">a federal law</a> that forces health care providers to give patients all of their information … for free; previously health care providers would lock this information up or charge high “processing fees” to reduce churn. What use-cases will be unlocked for patients now that they can get information from any doctor they visit and take it with them? On the provider side, current information request processes are inefficient, what problems will they face in the future?</p><h4>Societal Trends</h4><p>The more ambiguous type of inflection event is societal. This type requires analyzing changes in how people live, work, and play. Societal inflection events are identified by researching population trends like: access to high-speed internet, number of people working side-gigs, or number of people who fear climate disaster. While companies like <a href="https://www.notion.so/">Notion</a>, <a href="http://zoom.us/">Zoom</a> and <a href="http://slack.com/">Slack</a> seemed perfectly positioned for COVID-19 lockdowns, they were actually built for the increasing number of remote workers which has been <a href="https://www.ntia.gov/blog/2020/nearly-third-american-employees-worked-remotely-2019-ntia-data-show">doubling every year since 2012</a>.</p><p>So what trends are coming up outside of more remote workers? Our aging population! The number of people older than 65 years old went from <a href="https://www.urban.org/policy-centers/cross-center-initiatives/program-retirement-policy/projects/data-warehouse/what-future-holds/us-population-aging#:~:text=The%20number%20of%20Americans%20ages,reaching%2080%20million%20in%202040.&amp;text=By%202040%2C%20about%20one%20in,one%20in%20eight%20in%202000.">35M in 2000 to 50M in 2020 and projected 80M in 2040</a>. This increase has put a strain on the Medicare and Social Security programs, both of which will <a href="https://www.nytimes.com/2019/04/22/us/politics/social-security-medicare-insolvency.html">start costing more than their incomes in 2026 and 2034 </a>respectively (ie. 100% won’t be paid out due to lack of available funds). By 2029, it is expected that<a href="https://www.healthaffairs.org/doi/full/10.1377/hlthaff.2018.05233"> 54% of seniors won’t have enough money to cover their living/health expenses</a>. While elder care has been stagnant as an industry, it’s clear that the aging experience for this next batch of seniors will need to change! For example, companies like <a href="https://www.joinpapa.com/">Papa</a> have been working to make assisted living cheaper by creating on-demand assistants.</p><h3>How to identify opportunities using inflection events</h3><p>More often than not, an inflection event presents opportunities rather than exact answers. As entrepreneurs it is our job to look across industries, speak with real people, and consider how to leverage inflection events to create 10x better solutions. While your investigations should be fluid, I think that following the following 3 steps can get you pretty far:</p><ol><li>Keep track of inflection events — as an entrepreneur building the future, you should be constantly researching how the world is changing, internalizing how you think that future will look like, and connecting those predictions with problems you discover.</li><li>Talk to industry folks or customers about problems — you can’t avoid talking to people! Talk to friends but more importantly talk to people who have problems that you want to solve. Keep those conversations going to create opportunities for problems to connect with inflection events.</li><li>Predict a solution / Predict the future — <a href="https://www.georgesequeira.com/writing/go-from-nothing-to-funding-in-9-months-learn-out-loud">Think out loud</a>! Whether it be by formally pitching over the phone to a customer or writing emails to folks interested in the space, there is no way to know if you can predict correctly without trying to predict. Share your thoughts with industry folks, VCs, colleagues and most importantly: get feedback. From the entrepreneurs I talked with, you should aim for 50% of feedback being “that’s brilliant” and 50% “that’s impossible”.</li></ol><h3>Final Thoughts</h3><p>I’m super excited for companies solving the problems of tomorrow. My advice if you’re looking to build that future — do so through the lens of inflection events. Leverage societal, technological and regulation changes to be a part of the next wave of innovative companies. As always, speak to customers, colleagues, <a href="https://www.georgesequeira.com/about">me</a>, and industry leaders to validate your ideas: feedback being 50% “that’s brilliant” and 50% “that’s impossible” is the goal. Good luck friends!</p><p>—</p><p>Also published on my <a href="https://www.georgesequeira.com/writing/finding-startup-ideas-that-anticipate-the-future">personal blog</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a81086ad1680" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When building a company, learn out loud]]></title>
            <link>https://medium.com/@sequeira/go-from-nothing-to-funding-in-9-months-learn-out-loud-21be9bf89888?source=rss-268cc458acc7------2</link>
            <guid isPermaLink="false">https://medium.com/p/21be9bf89888</guid>
            <category><![CDATA[building-in-public]]></category>
            <category><![CDATA[collaboration]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[hustle]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[George Sequeira]]></dc:creator>
            <pubDate>Sun, 12 Sep 2021 20:25:53 GMT</pubDate>
            <atom:updated>2021-11-09T21:03:36.154Z</atom:updated>
            <content:encoded><![CDATA[<p>Also published at: <a href="https://www.georgesequeira.com/writing/go-from-nothing-to-funding-in-9-months-learn-out-loud">https://www.georgesequeira.com/writing/go-from-nothing-to-funding-in-9-months-learn-out-loud</a></p><p>As of September of 2021, I have stepped back from my role as Co-founder at RevvUp. While I rest, collect my thoughts, and prepare for my next adventure, I want to pass along some of the learnings from my journey.</p><p>By the start of 2020 I had helped launch the first Alexa product, built the foundations of a unicorn at Bluecore, and helped kick off Bonsai’s initial product. These were all incredible experiences that left me with an itch to create and grow a company of my own; there was just one problem… well three: I had no specific idea, no co-founder, and no funding.</p><p>For the first 9 months of 2020, I learned a lot that led to finding a cofounder, a product, and raising a $1.85M seed round. One of my most valuable learnings is the concept of learning out loud.</p><h3>What is learning out loud?</h3><p>Usually when people think of “learning”, they think of notes in Notion, Roam, a piece of printer paper on everything they could think of about a topic. As they collect these notes, hypotheses start to form and perhaps some conclusions reveal themselves. The thing is that up until the grand reveal, all the information is hidden from the world and people don’t know what you’re thinking. Learning out loud is pushing out these ideas (half-thought out is okay!) as quickly as possible to the world. These pieces of information can go onto twitter, blog, slack groups and if you’ve got a big enough network (10 people is enough), you can push them through your email newsletter. Be as noisy as possible. I’ll explain why.</p><h3>In the beginning…</h3><p>At the start of my journey, I knew nothing! I had just left my awesome job to start a company that did… nothing! I wanted to hide away until I came out with my genius idea. I felt like a term paper was due today and I had totally forgotten about it. I went into hyper-research and hyper-networking mode… in private.</p><p>I would go through cycles where I would go heads-down to learn about an opportunity and build a prototype, pull out my results to share with the world and use those responses to go back heads down. After 5 months of this, I had gotten that cycle down to 4 weeks. I was burnt out and as I watched my runway dwindle, I felt that this feedback cycle needed to speed up. What’s more, I was going at it one idea at a time. I needed to make this go faster!</p><h3>Opening up my research</h3><p>As I felt the shadow of time creeping up on me, I started to broadcast my ideas in real-time through emails, slacks, and at dinners, and not just the idea I was focused on, all of them, even my “Uber for dogs” ideas. To my surprise something amazing started happening. People were trying to help me out! Turns out when folks see you trying to figure out a problem, they jump in: I was getting in-bounds from people who had already tried an idea, introductions to folks who had a problem I identified, and even folks who wanted to invest if I pursued certain ideas.</p><p>I was learning a lot from folks based on my premises, killing ideas quickly, and identifying opportunities that I wasn’t even thinking of. The goal at the early-stage isn’t just to find a good idea, it’s about learning how to spot bad ideas. This scaled method of learning allowed me to spot similarities in bad ideas and unsuitable industries. By learning out loud, I had sped up my learning. By July of that year, I had gone through 20 ideas. What’s more, as I’d been learning and broadcasting out loud, folks got to learn what kinds of problems I liked, what values were driving me and my process in validating ideas: it started to create a funnel of potential co-founders.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZHzTbzdUD0cTZ8cbSLJr4w.png" /><figcaption>People interacting with your ideas ASAP is key in speeding up your learning</figcaption></figure><h3>Lesson</h3><p>For folks in the early-stages, there is an immense amount of pressure to learn a lot… fast. Learning out loud by sending emails, creating a newsletter, posting in LinkedIn/Twitter, and sharing as much in person, will give people an opportunity to supercharge your learning. What’s more you may find a co-founder along the way.</p><p>By the way, I’m currently exploring the connected fitness wellness space. If you are interested in this space and want to collaborate, feel free to reach out g@georgesequeira.com.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=21be9bf89888" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[On Exact Language]]></title>
            <link>https://medium.com/@sequeira/on-exact-language-205a0d9ae360?source=rss-268cc458acc7------2</link>
            <guid isPermaLink="false">https://medium.com/p/205a0d9ae360</guid>
            <dc:creator><![CDATA[George Sequeira]]></dc:creator>
            <pubDate>Thu, 29 Apr 2021 23:43:36 GMT</pubDate>
            <atom:updated>2021-04-29T23:43:36.610Z</atom:updated>
            <content:encoded><![CDATA[<p>Also hosted: <a href="https://www.georgesequeira.com/writing/on-exact-language">https://www.georgesequeira.com/writing/on-exact-language</a></p><p>Growing up as a child to immigrant parents, my role was to play translator in conversations that required english. Transacting on behalf of my family trained me to pick up on the nuances of language, how to sound “American”, and how to leverage colloquialisms. While it was annoying to deal with all the forms, phone calls, and meetings, it taught me that inexact and ambiguous language is where inherent biases thrive. Sometimes the “best deal” at T-Mobile came with hidden overages and oftentimes, “a table for 4” would put you in corners or near kitchens. These instances taught me that speaking loudly, clearly, and, most importantly; <strong>exactly,</strong> increased my chances of getting what I wanted.</p><p>Part of getting into STEM was the allure that numbers can’t lie. There is a right answer and a wrong answer, your program either does the thing or not, it handles the load or it doesn’t. So when I went into the tech industry I was excited to see this objectivity in action.</p><p>Unfortunately, I learned that inexact language screws people in the tech industry as well. Early in my career, a female co-worker got passed up for a promotion due to not providing enough “impact” for the team. While she mentored a few of us, extended our organization’s core platform, and designed exciting new features, the manager did not find her “impact” large enough. Who did he think provided enough “impact”? The guy who broke the build multiple times, never delivered on-time, and happened to get beers with the manager every other week. Her advice to me: get the exact milestones for your promotion and most importantly: what “impact” you can plan for. The same power dynamics in the real-world apply at work, even in the tech industry.</p><p>And it’s not just promotions and titles; Basecamp <a href="https://world.hey.com/jason/changes-at-basecamp-7f32afc5">recently announced </a>changes about their company policy barring employees from discussing societal + political issue talk (Coinbase <a href="https://blog.coinbase.com/coinbase-is-a-mission-focused-company-af882df8804">did so recently</a>). What constitutes societal talk? Does talking about the unsupported essential workers count? Does talking about the Black Lives Matter protest you went to count? It would be difficult to have lived on the planet this last year and not hear about some of these things. While they won’t make an exhaustive list of what you should and shouldn’t say, you can be sure that these rules will be enforced arbitrarily and in a biased fashion. Those in power will get to play favorites and moderate based on their bias’. This will leave minority voices more likely to not say anything at all, while those with main-stream or historically appropriate views to speak their mind.</p><p>Ambiguous language screws minorities over when paying for stuff, when getting paid, and even with how they are supposed to talk at work, what do we do with this?</p><p>We create and demand more exact language from the organizations we are a part of: work, social, fitness, coop, whatever. Ambiguous values, rules, principles could benefit from exact language, at least so that you know what weirdos not to hang out with. As someone who has dealt with this throughout my life, I’ve learned that shifting to more exact language benefits everyone. Humans are relationship-driven; clear communication creates closer relationships, less ambiguity creates less overhead in understanding, and clearer intentions builds trust.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=205a0d9ae360" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Estimating auto-scaled instance costs]]></title>
            <link>https://medium.com/bluecore-engineering/estimating-auto-scaled-instance-costs-9c598528f42?source=rss-268cc458acc7------2</link>
            <guid isPermaLink="false">https://medium.com/p/9c598528f42</guid>
            <category><![CDATA[bluecore]]></category>
            <category><![CDATA[google-cloud-platform]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[George Sequeira]]></dc:creator>
            <pubDate>Wed, 31 Jan 2018 13:01:01 GMT</pubDate>
            <atom:updated>2018-02-01T13:01:41.328Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>How can we reduce costs of running our code?</blockquote><p>Every engineer has answered this question at some point in his/her career. My latest encounter with the question was in early 2017. After two quarters of engineering heads-down building product, it was time to for us to assess the cost damage. Available at the time were daily billing reports provided by Google that we used to monitor costs and irregularities.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*94A04Q3TJPtlUeDh." /><figcaption>No we aren’t in the cryptocurrency game, Bluecoin is a proxy so we don’t expose the dollars.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*P7lzaBMtsTLrRsDu." /></figure><p>These reports were too high-level to derive optimization insights; however, they highlighted that most of our costs were due to auto-scaled instances. At Bluecore, every squad is responsible for a set of APIs, most of which run on these auto-scaled instances. As such, we weren’t surprised to find a large portion of costs were associated with them. In order to empower squads to investigate possible optimizations, we needed cost reporting at the API level. This meant that we needed to answer:</p><blockquote><strong>How much does it cost to run my code?</strong></blockquote><p>In this post we’ll step through some of the concepts about AppEngine’s auto-scaled instances, why it’s difficult to calculate costs for them and how we mixed various sources of data to provide good estimates on API-level costs.</p><h3>Auto-Scaled Instances</h3><p>The instance is the basic building block in AppEngine. It is the underlying server receiving and handling requests. AppEngine abstracts out management — starting/stopping/scaling — of instances behind a service. A service is distinguished by what API routes it handles and has configuration for:</p><ul><li>Template for instances handling requests (e.g. 1GB memory)</li><li>Parameters to scale the instances up and down (e.g. max 50 concurrent requests)</li></ul><p>For example, let’s take our event ingestion service. It handles one specific API: <strong>/api/integration/event_ingestion</strong>. This service is tuned to have low-latency (&lt;50ms) and as such we reduce the amount of concurrent connections and turn up the number of idle instances.</p><p>AppEngine automatically scales this service for us so that we don’t worry about the infrastructure. The following graph shows the number of requests handled by the event ingestion service compared to the number of instances running.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*mAdG2-yN3cTTyZTB." /></figure><p>More importantly, the 95th percentile latency consistently stayed below 50 ms in this time period:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*MphLyS3EdrHsbXRF." /></figure><h3>Cost of Instances</h3><p>Google only bills for active instances and does so on a per-minute basis. An instance’s rate depends on it’s CPU/Memory capacity. For example, an F4 instance (512 MB memory 2.4 GHZ CPU) costs $0.20 per minute and an F2 instance (256 MB memory, 1.2 GHZ CPU) costs $0.10 per minute.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*XtEELnM9CpWtjdg5." /><figcaption>We only see the yellow graph, yet we want to dig into the F4/F2 graphs as well.</figcaption></figure><p>That being said, as these instances are scaled up and down throughout the day, it is difficult to say for sure how many instances are used as it depends on the traffic being handled. Revisiting the prior graph, you can imagine what that might look like:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Sjh_ok3eaxknrVZ7." /><figcaption>Unpredictable costs if the F2s are scaled independently. Imagine trying to parse the instances from 25 services!</figcaption></figure><h3>APIs and Costs</h3><p>Early in Bluecore history, engineers found that many APIs shared similar instance requirements (most run fine on F4s for example). To reduce the burden of managing/deploying too many services, we grouped APIs by their operational requirements (CPU/Memory). At the start of 2017, we had 231 APIs managed by 25 services. This meant that even if we could calculate the number of billed instances per service, we would not be able to directly tie it to an API.</p><p>The following is a graph of requests handled by one of our F4 services:</p><p><strong>Number of requests handled</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*aHkNs-Pv9QnSRd0k." /></figure><p>This service has over 50 APIs that vary in their CPU utilization and latency profiles. We needed a way to properly attribute instance utilization.</p><h3>Logs to the Rescue</h3><p>In order for us to provide API-level costs, we knew we’d need to dive into the request logs to stitch together information. AppEngine’s request logs are rich with information such as:</p><ul><li>Service Id</li><li>Response code status</li><li>URL Path</li><li>Referrer</li><li>Latency</li></ul><p>At first, we thought we could simply use request times to estimate costs (the sum of all request times should be the amount of instance time spent processing the request), but that approach omitted the multi-threading feature of AppEngine (one instance can handle up to 50 requests).</p><p>Using only latency also had the problem of neglecting the different rates per instance type (F4/F2/F1/etc.): we would report that a one-second request on a regular F2 was cost-equivalent to a one-second request on a beefy F4. We wanted the F4 to be reported as 2x that of the F2 to<a href="https://cloud.google.com/appengine/pricing#standard_instance_pricing"> reflect the rate difference</a>. While looking in the logs for other metrics to leverage, we found “megacycle” — an integer that “corresponds to 1/1200 CPU second” (<a href="https://stackoverflow.com/questions/5213576/appengine-megacycles-to-cost">stackoverflow post</a>). After experimentation, we observed that for a one-second request, the megacycle difference between an F4 and an F2 were 2 to 1 respectively.</p><p>To show this, I deployed a new API to an F2 and an F4 instance called <strong>/api/basic/demo_megacycles</strong> that prints an incrementing counter in the logs until 500 ms have passed. The endpoint simply runs the following python method:</p><pre>import time</pre><pre>def execute(self):<br>  start = time.time()<br>  i = 0</pre><pre>  while time.time() — start &lt; 0.5:<br>    print %s’ % (i + 2)<br>    i = i +1</pre><p>The megacycles reported were:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_MAzH00FNk7qTByvuHBaJA.png" /></figure><p>The megacycles line up very close to 2x different!</p><p>When we investigated using megacycles as a proxy for costs, our testing across similarly behaved APIs gave us intuitive values. It’s easy to <a href="https://en.wiktionary.org/wiki/yak_shaving">spend too much time</a> finding a <strong>perfect</strong> measurement, but we found “close-enough” allowed us to start providing value to the organization faster. When you find yourself trying to measure costs, don’t be afraid to ground your costs to something that isn’t perfect. Get data that <a href="https://en.wikipedia.org/wiki/Minimum_viable_product">enables action</a> and then iterate on your measurements.</p><blockquote><strong>Ground your cost (imperfection is okay!)</strong></blockquote><p>We needed access to these logs so that we could sum megacycles by API. Therefore, we <a href="https://cloud.google.com/logging/docs/export/configure_export_v2">exported our logs</a> to BigQuery. A daily process would query BigQuery, extract megacycle and API data and send it to our analytics platform. Here is a look at the top 10 APIs in our megacycles-per-API graph:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*eZh_qlKv8MtcvaDI." /></figure><h3>Tying it All Together</h3><p>At this point, we had our daily costs for auto-scaled instances and a per-API megacycle count. In our analytics platform we then divided the daily auto-scale instance cost by the sum of accounted megacycles. With this rate, we could provide per API costs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*R2_KqRkFlLzS849w." /><figcaption>Once again, we use Bluecoins as a means to obfuscate cost.</figcaption></figure><blockquote><strong>Measure, hypothesize and validate cost optimizations.</strong></blockquote><p>This cost graph has been instrumental in allowing us to measure, hypothesize and validate cost optimizations. Unexpectedly, they have been useful in identifying costly bugs as well. In March 2017, we found an API had a memory management bug by using the cost dashboard. The API would sporadically eat up too much memory and force more instances to scale up more than expected.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/942/0*9d_aA9beZZfQm7kM." /></figure><p>With this dashboard we were able to identify the issue, hypothesize a fix and validate that solution. The per-API dashboard has increased our engineers’ understanding of their code and empowered them to manage costs respectively. In the future, we would like to add automated alerts to catch situations like the bug described earlier.</p><p>If you find yourself needing to optimize your codes’ cost, you’ll need to:</p><ol><li>Find out what your code costs to run</li><li>Ground your cost model (perfection optional)</li><li>Measure, hypothesize, validate</li></ol><p>Also, if you find yourself looking to work somewhere that lets you deploy APIs in a matter of minutes and have pure ownership of your code, drop us a line ;)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9c598528f42" width="1" height="1" alt=""><hr><p><a href="https://medium.com/bluecore-engineering/estimating-auto-scaled-instance-costs-9c598528f42">Estimating auto-scaled instance costs</a> was originally published in <a href="https://medium.com/bluecore-engineering">Bluecore Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>