 Hello, hello, good morning, good afternoon, good evening, wherever you are in the world. My name is Anand Deshvane, I'm the head of the startup program here at Twilio Segment and I'm very excited to be here with you and to talk about how to use analytics for startups in the early days and how to set up your first stack as well. We're going to be talking about a lot today and so I'm excited to work with you to help you break down some of the silly logistical barriers that you might approach as you're entering into the first days of your startup's existence. A lot of what I'm going to be talking about is going to be guided by the story of these four guys. These are our four co-founders of Segment when they were in summer of 2012, a batch of Y Combinator. They just dropped out of MIT and they were working on a live lecturing tool called Classmetrics. So the Classmetrics was a live lecture tool that allowed students to say whether they were confused or not during a class and the professor or the TA would get a live cardiogram of the class's confusion. The idea was that the professor and the teacher could get an understanding of what was going on in real time and could stop the class and help the students in real time too. Now, not a huge surprise now but unfortunately it turned out that the only products used adoption they were increasing were Facebook and Amazon. Professors also didn't really seem to care so much about whether students are confused or not during the actual lecture and students just don't stay on educational sites for very long. So as they were trying and failing at figuring out how users use Classmetrics, they decided to build an analytics product. So they built one and they built another one and they ended up building five more and the only thing that was really similar between any of these products was that no one really wanted any of them. Each time they made a new product they were missing the larger picture. They were going about product market fit completely wrong. So December of 2012 was a very low point for all four of the founders. They had built a bunch of products, publicly launched zero of them. They had no real customers and they're running out of cash with about seven months of runway left. They were all completely burnt out and so they met with Paul Graham who's a wide commonator. He pretty much summed it up to the founders this way. So you've burned through half a million dollars and you're back to square one again. Yeah, I would not like to be those guys. So as they were trying and failing at getting Classmetric off the ground, they were learning about analytics for the first time. And in the screenshot, our co-founder Peter is trying to integrate analytics into our code base for the first time. In general, they were incredibly confused by which tools to use, which metrics to use and look at and how to integrate all of these tools. They spent way too much time on this problem. Specifically here they're struggling with the amount of time each integration took them. So then they stumbled onto an idea. What if they implement their analytics with a single simple API? That API would be as simple as possible. Who is the user and what are they doing? These two API calls serve as the basis of segments tracking API even today. If you instrument your tools with this API, you can add any analytics and marketing provider without writing any additional code. So this is what our founders realized that they should have done. They should have taken their early experiments to the market earlier to understand if there was a real need before for the pursuing each idea. They had a breakthrough on December 12, 2012. All the time that they were building these analytics tools, the answer was right under their noses. They had an open source analytics library that helped companies instrument analytics tools with a flip of a switch. As a Hail Mary, they decided to post up a landing page on Hacker News, which is one of the most popular developer forums and the experiment ended up working. Their open source library went to the top of Hacker News. Within 24 hours, they had thousands of signups, tweets, and comments about people wanting this at work. This is what product market fit looks like when customers start pulling instead of you pushing. So riding that high, they spent the next two weeks furiously coding and they shipped the server side version of segment almost exactly the way it looks today. You can send data to segment, flip a switch for Google Analytics, and the data would magically show up in GA. And over eight years, we've built an entire company on top of that initial Hacker News launch. Over 20,000 customers across 70 different countries, but they could have gotten there so much earlier and here's what they should have done in the early days. Before we jump into it, segment is a much larger product today. We can read data from just about any data source. Your website, your mobile apps, your server applications, CRM data from Salesforce, payment data from Stripe, marketing data from HubSpot. That collected data then flows into Segment's platform where it's unified into user records. We then allow you to integrate with 350 different tools, including analytics and marketing automation tools, as well as raw data sources, like Amazon S3 and data warehouses. So over eight years, we have been very lucky to catch almost 20,000 different companies of all different sizes and types. We've caught companies like New Relic, Instacart, Deliveroo, and learned a lot of lessons while working with them. I'm here to share insights about how companies tend to evolve over times with their initial analytics needs. And the answer really is analytics. It's a simple answer, but if you get in on day one, you're much less likely to waste time and money chasing fake ideas. So here's how you should use analytics and data in the early days. So we are going to be focusing on these top three reasons of why startups might fail. This is based on third-party research by CVI. It's a much bigger topic that has a lot that we could get into, but for the sake of making this kind of easy to digest, we're going to be focusing specifically on these top three reasons, helping you find the product market fit using your analytics, helping you to raise funds using appropriate analytics and metrics, and making sure that you're guiding your team in the right direction. We can use data to help you find your right team, but we won't be getting into that today. And I'm going to go over what we're going to be covering today. As I mentioned before, I run the startup program here at Segment. I've been at Segment for four and a half years, been running the startup program since it launched about two and a half years ago. And I have had the pleasure of working with thousands of startups personally, you know, really working with probably hundreds of startups at this point. So I'm going to be using that along with my mentoring. I mentor at 500 startups at GSV Labs, now called One Valley, a number of other places and I'm an advisor and angel investor too. So I'm going to be using a lot of what I've learned from the folks that I've worked with so far, as well as kind of tribal knowledge aggregated through years of work with startups at Segment generally. So today we're going to be starting with the basics, setting up your first funnel, collecting your first data and designing your first metrics. This section is designed for getting started with analytics. So if you already have metric set up, stay with me and we'll get into more advanced stuff in the next section. Next, we're going to cover the analytics and growth tools that our best customers use that Segment uses and that I personally recommend. Then we'll dive into how to do data collection the right way. After that, we'll talk about what mistakes you should avoid. We've seen some real horror stories on the Segment platform and I want to make sure that we pass those common mistakes on to you so that you don't have to go through the trouble of making them. Then we will go into the startup stack. So that's our founders' personal recommendations of what tools startups should use. Then we'll leave you with what you should do today. Okay, let's jump in. So creating a funnel is the first step in setting up your analytics. Funnels help you model your business as a series of steps that customers experience in order to achieve value from your product. We'll use Netflix's funnel here as an example. Now, don't worry if you are not a B2C company. If you're a B2B company or a B2C company, this is the basic funnel for every product for both of you. Mobile and e-commerce, you can also use this too with some very obvious shifts that you might have to make. These are the three key steps that you need to look at. Acquisition, engagement, and monetization. Retention is when your users engage over longer periods of time. Again, we're focusing on Netflix here because Netflix is easy to understand. Everyone knows their business model. Then you ask the three core questions around the funnel. How many signups did I get this week versus last? How many users were engaged week over week? And how much revenue are we making this week versus last? And then you customize the funnel to your own business. This is a simplified version of Netflix with a free trial starting when the user signs up. Excellent. Now let's go ahead and talk about how to collect data for this funnel. First thing is to collect your data. You'll want to instrument your app with analytics.track calls. An analytics track call ties the user ID, the who, to an action, the what. You can use a string to represent the action so you can tie it to your own business. Use event properties like signup type or video ID to describe what type of video was played or subscription was upgraded. Every event can have event properties, and event properties allow you to ask deeper questions around the event in your analytics tool. Some examples of these questions are what video category is most popular this week, what video category is often watched before someone purchases a paid subscription, how much monthly recurring revenue are we making, and so forth. Now the event property I like to think of as kind of the what as well as the how. And so it's really useful. You're going to be tempted in the early days to track everything, to track a bunch of different events. The better option is actually to track a limited number of events with really robust event properties associated with them. Awesome. So next, you can see all of your data in the second debugger. The debugger will scroll automatically as new events come in. You can see the exact payload of what you're sending into segment in real time within the debugger. It's a really important part of our tool. Now it's time to add your first analytics tool. So sign up for Mixpanel and grab a project token that's the API key in Mixpanel speak. Next, go to press the add destination button within your segment workspace and add Mixpanel using its token or its API key. And that's really it. Your data will immediately start to flow into Mixpanel and here's a Mixpanel event graph of user signups and sign ins. We can really easily see how they're growing over time. So one thing that I'll mention here is I'm going to be talking about Mixpanel quite a bit during this. It's not necessarily product placement. If you prefer to use TARDEO or prefer to use amplitude or if you want to go ahead and use a data warehouse and ETL through segment into that data warehouse and put a BI tool on top of it, go for it. No harm, no foul. I'm talking about Mixpanel here just because anyone who's part of our segment startup program, which I'm going to talk about in a couple of minutes, gets Mixpanel for free for two years. And so we want to be able to offer that to you. So that way you can use this information and get started immediately. If you prefer TARDEO or amplitude, go for it. You can use all of these tips and tricks with those tools. Now, if you have a BI tool setting on top of a data warehouse, you might have to make a couple of adjustments because you're building it from scratch, but you can still use all of this information, I promise. So next, we're going to focus on three metrics. This is going to be very, very important. Now, Mixpanel will graph signups per week. This is a graph that we grouped by the signup type and by the attribution channel. So we can see here how many users a week are invited users versus organic users, but we can drill down a little bit further to be more specific. So first off, go ahead and create an insights report in Mixpanel and select your event. Here we're going to choose the event signup and then we're going to start breaking that event down by other factors like the platform that's iOS versus Android versus web and the attribution channel for your signup. So that's the source of your signup. Now, Mixpanel will now show us signups by platform that's web versus mobile and the attribution channel. In this graph, we are grouping by the platform and the source. We can see here how many users in the past week are on each platform. That's mobile versus web and are from what channel, organic, invited, Facebook, et cetera. And that's it. Now let's move on to retention cohorts. So next up is retention. This helps us understand whether users are getting recurring value in the product over longer periods of time. This chart is a little bit more complicated, but it's incredibly important. So you can see here that the retention rate for different platforms by day. So that's what percent were retained and what percent were dropped. And easily create a cohort to take action with those users. So Gustav Alstomer is a partner at YC and was previously growth product lead at Airbnb. A couple of years ago, he gave a talk about how to pick metrics that matter. In Gustav's talk, he talked about finding one metric that represents value for your customers and then finding a frequency within which you will measure that value. So for Airbnb, for example, they are optimizing for annual bookings per customer. For Facebook, they are optimizing for monthly active users and observing this metric over time is incredibly important for understanding whether you have product market fit or not. So what we've seen is that for companies to really measure success in addition to having summary KPI like the ones I just showed, they also have growth KPI to better understand how many of their best users they are getting over time. You probably heard of Facebook's famous example of understanding that when users reach their magical milestone of getting seven friends in 10 days, then they were more likely to stay and be engaged on Facebook. Now, there are many mistakes that people make with assuming causation for metrics that are really just correlations and you have to test out different factors to assess this out. I wish I had a little bit better of a tip to give you for this, but honestly, you really do have to play around with it. If you have a data analyst or a data scientist friend who is either on your team or willing to work with you, those folks are magic when it comes to figuring this stuff out. Another thing is that I know that there was a tool called ClearBrain that was really useful for this. It was acquired last year, I believe, by Amplitude. I'm not sure exactly what they've done with it, but perhaps there's a tool there for you to use. The idea of learning what makes your best users stick around early is extremely helpful to galvanize your teams to focus on key growth KPI and continually work to drive it up. Now, the growth KPI can act as a leading indicator to engagement and overall retention, which is really important when times of the essence in the life of an early-stage startup. I personally think that growth KPI, key growth KPI, that use your key retaining event, but finding specifically your best users and how they're sticking around and how that number is affected by any experiments you're running and product shifts that you're making, that's probably one of the most important things that is not talked about enough for startups. Key growth KPI are just so important, I think that you should all really be looking at it and you should reevaluate it every six months or so as your company starts to shift, as you start to learn more about your users and as your product pivots even. So the overall basic idea here is that products with product market fit will see users continue getting value over longer periods time whereas products with no product market fit will see their users stop using the products within a few weeks or months leading to high churn and indicating no product market fit. That's why it's important to define key growth KPI early to redefine them somewhat often on a medium to long-term basis and to continue to look at them, to hold them near and dear as you're looking at your experiments. So let's take a look at retention a little bit more. So this is quite a simple idea and it's game-changing. It's also something that I see a lot of founders that I work with um simply not having a good understanding of or forget about. It's also something that our own co-founders didn't have early on. So if you don't have product market fit simply put users are not returning to use your product again. This is a cohorted retention chart. It's used to determine whether a group of users will be retained over time. This diagram follows a key cohort of let's say 100 users who signed up during a specific time period as they use the product for many weeks since they signed up. Each week we track how many of them perform the key target engagement event. Products with high product market fit will see more and more users from the cohort start to get value. Products with no product market fit will see the engagement percentage of the cohort drop to zero. So it's really easy to create retention cohorts in Mixpanel. First we select our key retaining event like watching a video and Mixpanel will bucket users into day, week, or month cohorts. We can then segment this cohort by breaking them out by the platform that they use, web versus mobile. This gives us better groupings within retention rather than just arbitrary days. So then we just click on a day box. Here we're clicking on day seven and Mixpanel lets us make a retention cohort in just one click, either for users retained on this day or users dropped on this day. It's up to you whether you want to see it glass half empty or glass half full. So now that we've built a cohort for retained or dropped users, Mixpanel will now show us the membership or the size of this cohort as it changes over time. For example, we had a huge spike in seven day iOS retention on the week of March, let us say 11 to 17, which is about a 66% boost in retained users. Now this type of growth would be very hard to observe without the ability to see how cohort membership changes over time. Now this isn't just both users changing over time, this is a specific cohort. So this means that the video content that we published this week had a massively impactful effect on our retention, which sustained itself into the next week, but then trended down over time. So Mixpanel can also visualize our revenue over time tracked as a property inside the insights report because of the way that we set up segment tracking at the beginning. So whether it's through in-app purchases or subscriptions. Now here we're showing the total revenue from our application day over day and benchmarking it against the previous month, which helps us to see if our revenue is up or down relative to the previous time period. Now let's see, we can also frame our in-app purchases or subscription revenue in the context of cohorts. In this example we'll look at total revenue from in-app purchases for users in the new loyalty program and all users cohorts and further break this out by the date to see any differences in spending patterns across the week. Mixpanel's insights report tool makes it easy to query this data because we already have the component cohorts events and user properties to slice our data. We know our incentivized users generate most of our revenue, but here we find a high amount of revenue coming from those users who are in our new loyalty program shoppers. So Mixpanel's insights report also makes it easy to build this query out because we already have the component cohorts because of the way that we've set up tracking within segment. We also know our business accounts generate most of our revenue, but here we can find a surprisingly high amount of revenue coming from our creators who are in our seven-day retained iOS cohort. So and this is because of the way that we've set up Segment and Mixpanel together. Excellent. So now it's time since we have all of our metrics set up to look at our first dashboard. Frankly, everything that we've been doing so far has been in service to setting up our first dashboard. Simply put, our dashboard is how you set goals for your teams and set up accountability around those goals. Simply put, our dashboard is how we operate our company. Mixpanel will let you drag existing charts into a single dashboard that you can share with your company. You can even set up goal targets and watch whether you're meeting those goals or not. Our goal for seven-day retained iOS users, for example, is 4,000. Now the best dashboards are like products. They have engagement and retention, and if you build a dashboard that goes stale and no one looks at it, then it was a waste of time. The best dashboards are the sorts of truth for everyone at the company. They're shared around and obsessed over, and they're used at every major company meeting and every company all hands. So Mixpanel also has a mobile companion app where you can see your dashboards and metrics on the go as well. Now, one other tip that I'll give you is that you should send a monthly email to your investors and advisors with an update of your week over week metrics. It's a great opportunity to report how you're doing and ask for help. This allows your advisors to give you targeted relevant advice. The best companies I see are doing this. We have done this at Segment ourselves. It also allows your investors to show you off more easily. The reason I say that is because we've already gotten four different accounts from people within the Segment startup program that have said that because they were able to use Segment from the early days, they were able to set up their metrics early and were able to report early, and their investors were actually circulating some of their successes, and were able to get inbound lead rounds through that circulation of their successes through their investors. One of those folks specifically, they would post their week over week metrics on their Friday weekly roundup blog, and one of their investors was screen shotting every week from the blog and posting it on Twitter. And that investors friends were seeing the week over week progress through Twitter and inquired about their next round. We wrote a customer case study about them. It's in our customer stories. I believe that was the parable case study. If it wasn't parable, I believe it was help. But one of those two stories talks about this. So not promising that that will happen to you, but if you get it started early with your analytics, then you have a chance for it to happen. So now that you know what metrics to build, you'll need a few tools to set those metrics up. We see a lot of founders have analysis of what tools are best. This section is our recommendation of what's best in class in the market today based on our own experience. So naturally, your needs will evolve over time. This is the graph showing what tools are needed at each stage of your business. Do not worry about choosing the tools that will last a lifetime. Choose what's right at your stage and build in the flexibility to switch. That's a big bonus of integrating using segment is that you are married to your data, not to your tools. Anytime someone new joins on to your team and wants to build out something in their stack, they don't have to bother an engineer. You already have segment set up. They can go ahead and self-serve, switching on and off any tools that they want to add or deprecate, running their own proof of concepts, and doing trials without ever having to bother any technical resources on your team. So here is a graph of when you see different startups adopting various tool categories. The stages are building an MVP, releasing that MVP to a small group of test customers, then launching, then searching for product market fit, and then finally scaling the business. The first thing that people get is web and event analytics to understand whether customers are acquiring or engaging. The founders usually set this up before giving the product to their first customers. A cool little hack is then to put in live chat in the app to capture any customer feedback that comes your way. That way you can marry some of your qualitative data with the quantitative data that you're already getting from web analytics. If you already have a ton of data or want to ask complex questions, you can also throw that data into a data warehouse like Redshift or BigQuery and then query against it using a BI tool. Segment works as an ETL pipeline for these types of warehouses, so it's really easy to do that if that's what you want to do. I will say that I'm a big fan of having a data warehouse and a BI tool. I think every company should have one eventually. I do not believe that most early stage startups need one. It's drinking out of the fire hose. If you have less than 10 people on your team, you probably shouldn't get a data warehouse just yet. Awesome. Next, founders will install a company dashboard to focus on the right metrics to grow the business and then we'll push an email providers to send the first emails to customers and then it helped us to answer any tickets that are coming in from those customers. Once the company has found product market fit and is starting to scale, a CRM is set up to grow a sales team and then advertising is added in order to help fuel that sales team. Now, I'm going to go ahead and give you three different tools and give you a little recipe around each of those tools that will be helpful in the early days. So one common pattern that we see is that early stage startups have a group of technical co-founders and a group of non-technical co-founders. The non-technical co-founders ask business-minded questions and the technical co-founders are querying Mongo or MySQL database to find the answers. How many users do we have on the sites, et cetera? So that's an example of a company that does not have democratized data. In a company that has democratized data, the non-technical co-founders are also able to query the data to find their own answers to the questions that they have. We recommend that non-technical co-founders learn SQL so that they can query their own answers. You can use Segment to send your data from sources directly to Google BigQuery, Redshift, Snowflake, and then use a BI tool like Mode or Looker to be able to query against that data. We have exactly that set up here at Segment and you can see here where we run a query to see which of our integrations are most commonly enabled. Google Analytics, unsurprisingly, is the top one. At Segment we talk about how if you don't know SQL when you join Segment within six months, you will be pretty proficient. I personally ask entirely too many questions so I was proficient within the first three months of getting to Segment. So any new product is riddled with usability issues for about three months. This is a story specifically of SQL traits, which is part of Persona's product that we shipped about two and a half years ago. When we launched it in the first week, we saw that the metrics just were not good. People weren't making it all the way through our setup process. One of our designers had the idea that we should use full story to check out what's happening in app. We checked out the activity of one of our users. His name is Kyle here. Here we can see that Kyle goes in and tries to use the product. Kyle runs into issues not knowing if this button was actually saving the work, if they were progressing through this workflow or not. And in frustration, Kyle just quits. And what we realized was all we had to do was fix that one button to make it more obvious that it was working, that they weren't moving through the steps appropriately. And as soon as we did that, we saw a huge uptick in use of this product. It's now the most popular part of our most popular product on our platform today. I personally use this at least weekly and I love it. So next is the 43-minute founder email. Since 2013, we've had this email going out to all of our signups. We wait about 43 minutes and send this out through customer IO automatically. Since it's inception, we've had over 100,000 responses to it and our support team answers these emails as soon as they come in. We wait 43 minutes here to make it feel a little bit less automated, although obviously it's a design email. It's not coming from our co-founder. However, 43 minutes also gives people the opportunity to start self-serving. You are teaching people how to go ahead and learn how to use your knowledge hub. That way you're reducing tickets down the line as well. It's better for you. It's better for them. Awesome. So founders always ask what tools we recommend and this is our set of recommendations. I'm going to go ahead and go from bottom to top, going from left to right where appropriate. So for data infrastructure, we do obviously recommend Segment. Even if I didn't work at Segment, I would definitely recommend it still. Segment is super easy to use tool. It is really easy to get set up. You can add it into the header of your code and easily pipe data to any tools that you want. It's really great for early-stage startups because no longer do you have to spend any engineering hours on building infrastructure. Your engineer signed up to build your product, not to build pipelines. I personally don't come from a technical background as might be obvious, by the way, that I describe things. So when I joined Segment, I set up a video camera and recorded myself setting up Segment. The entire video consists of me adding the Segment snippet to the header of my personal website, piping the data through Segment, seeing it in the debugger, and then adding it to a mixed panel and a group GA destination, and seeing that data flowing into both of those. It's a very simple version of the setup, but nonetheless, it was the full setup. I saw the data piped. That entire video is 16 minutes long, and I swear it was just that easy. Now, of course, if you have a more complex setup, it might be a little bit longer, but it won't take you a super long time. So for analytics, Google Analytics, if you don't have it already, you should go ahead and have it. If you have an MVP up and running, Google Analytics needs to be there already. It gets you almost all of the way to the end goal, and it's free. It's a no-brainer. Mixpanel is the next thing we recommend on top of it, and as you start getting deeper into your data on Amazon Redshift with a BI tool, like Looker on top of it, and you can use Segment as the ETL pipeline for that. For customers, Intercom is a great way to keep tabs. For our experiences, we just talked about full story for messaging, customer IO, and for growth, nothing beats Google and Facebook ads. Now, we also talked to our own founders about where they were cutting costs and where they were spending in the early days of Segment. One of the things that you'll see really is that we paid for things that were core to our product, and we cut costs on things that were more GTM or product-focused, product management-focused. I suspect also some of these tools just didn't exist yet or didn't have a free version. Excellent. Now that we have talked about all of these things, let's do a little review and talk about what you should do once you get off of this call. First off, make a plan. Create a tracking plan with your object action framework. What I mean by that is go ahead and visit our website or our docs. There are a bunch of tracking plan templates that are based on different business plans, e-commerce, B2B SaaS, mobile, B2C, so on and so forth. Use those templates and create a tracking plan. Go ahead and integrate Segment. Track your data. Send data to Segment, and I'm going to show you how to get Segment for free in just a second. Then pick your tools and integrate with create tools like Mixpanel and a bunch of the other ones that I talked about. I'll also show you how to get those tools for free shortly. Build out your metrics. Start to monitor your three critical metrics, signups, retention cohorts, revenue, so on, and distribute. Build your first dashboard, and email your investors. Start to use your dashboards at your company all hands, so on and so forth. Segment is now free for any early-stage startups that are less than two years since founded, less than $5 million in institutional funding, and are independent startups. You cannot be a spin-off. You cannot be a subsidiary. You get $25,000 in Segment credits each year for up to two years on the monthly team plan. You get over a million dollars in other software deals from our partners, which I'm going to show you in just a second, and you get a Segment discount once you graduate off of those two years, which lasts for two more years. You could spend four full years basically getting Segment for free for most of it and then for really cheap for the rest of it. Included in our startup program are a bunch of partner tools. Of the tools that we already talked about, Segment, Google Analytics is always free, but Mixpanel, Full Story, and Customer IO are all free. We have a bunch of other ones that are free for at least one full year. Zendesk, Redshift, Zendesk, Adjust, Hot Jar, a bunch of them are all in there. And then Amazon Redshift, AWS gives up to $25,000 in credits. Look for a 70% off. Intercom, I believe, is 70 or 80% off. A bunch of other ones are minimum 60% off. We have almost 50 different deals as part of our startup program. If you enjoyed this session, please go to bit.ly-tweet-segment.oh for office hours. It would be great to have a holler from you about this session. We'd love to get that love. If you want to join our Segment startup program, please go to bit.ly-segment-oh, and you will be able to apply for our Segment startup program and get all of those benefits that we just talked about. It has been an absolute pleasure to be here with you to help you to figure out how to break down some of these logistical barriers to making your startups successful, using analytics. Really excited to see you folks take off in rocket ship. Wherever you are in the world, please be safe. And if you can't be safe, please be careful. There's still a pandemic out there. I hope that you have lots of luck in your journeys and we're really looking forward to growing with you. Have a great rest of your day, week, morning, wherever you are.