YC Partners Anu Hariharan and Adora Cheung answer questions from Startup School founders on how investors measure startups.
Thank you all for having me, and it’s so awesome to see so many of you at 9:00 A.M. In the morning to discuss metrics. Let’s hope we keep you engaged until the end of the session. How do we think about what metrics to track? Our advice is find out which business model you fit in.
The most common thing people do is which industry or [inaudible 00:00:24]. Are you health care? Are you biotech or are you enterprise? But that’s not really the best way to think of metrics. The best way to think of metrics is how do you plan to charge your users, which is the business model? And which of these business models do you fit in? So roughly there are nine business models. I would say 99% of you should fit into one of these categories. If you don’t, you’re probably building something that’s incredibly hard, which is what we call moonshot. So I’ll walk you through for the rest of the presentation on each business model and what three or four metrics you need to track. Beyond three to four, honestly, at this stage is an overkill. So these are the things that would matter.
So what is an enterprise business model? This is a company that sells software or services to a large enterprise. Pretty simple. Very few startups do that. So I would imagine very few of you try to or planning to launch something from day one that sells to say Facebook or Google or Apple or any of them. But if you are one of those companies, examples that are like Docker, Cloudera, FireEye, even in the YC portfolio there are very few that did that from day one. But if you’re one of them that sells to large enterprises, you would categorize yourself in this category. And the large enterprises tend to work in terms of contracts. So your business model will come into three things, which are the three metrics you track, which is bookings. So if you’re working with say Facebook, they’d say, “Hey, you are going to help us hire x engineers. We’d like to sign $a hundred thousand dollar contract for next year.” So that’s why you’d say, “What is my booking? What’s the total number of unique customers I have? And what’s revenue?”
The difference between booking and revenue is Facebook might sign a contract ahead of time, would tell you at the start of the year that it’s a hundred thousand for x hires over the year. That doesn’t mean you recognize revenue straight away. You’d recognize revenue only when you have delivered the service, either you’ve placed all the hires that you’ve set in your contract or if it’s an annual contract, you just divide it monthly. So the common mistakes we see founders do is confusing bookings and revenue. They would have signed contracts but they haven’t delivered anything. The contract hasn’t even kicked in but they’re already reporting it as revenue. That’s not true because you haven’t delivered service. So therefore it’s not revenue. And so you should hold yourself accountable for that like the company is not generating revenue.
The second common mistake, which is probably more relevant at the stage that you’re in is Facebook might have verbally told you, “I will consider a hundred K contract, for them a hundred K’s not a big deal for a lot of you, it’s a really big deal. That’s neither booking, not revenue because it’s a well built offer. So the other than, even if they signed a letter of intent, it means nothing. So you really have to have the contract written down, signed for it to be bookings and for revenue only when you start delivering it. The second business model is SaaS, which is probably where most of you fit in, especially if you’re in B2B or servicing to other companies. You’re all probably thinking of SaaS model. It’s a very predominant model these days. You see a number of all these other YC startups Segment, Ironclad, Sandberg, they all started with SaaS model, right from day one. SaaS is Software as a Service in terms of business model. It’s really subscription business.
You charge something monthly for a software that you provide. So what are the four key metrics you would want to track? Well, if it’s subscription, by definition, the revenue is recurring, right? Which means if you hopefully have built something that people really like, they’ll continue to use it and they pay you every month. So that’s why you track MRR at the highest level that is, Monthly Recurring Revenue. How much are you making monthly and what did the customers come at to it? ARR is Annual Recurring Revenue. Because of this phase, you’re really growing fast. You may be growing 30% week over week or 40% week over week. So it’s helpful to track your annual recurring revenue as literally run rate. So if this month you made 30,000 in MRR and if it’s truly recurring, you should expect that your annual recurring is 12 times that. So it’s just a good metric to use internally as well because it helps show the pace rather than just looking at absolute.
The third thing you should pay attention to is chart. So when you launch, there’ll be few users who are early adopters that use it, but you really should be paying attention to if they stop using it, right? And if you’re a subscription business, the chances are you probably make 5,000 to $10,000 a month from two or three customers, or maybe even 10 so losing a customer means real impact on your revenue. And so this is why we recommend measuring something called Gross MRR Churn, which is say at the start of the month, you expected your monthly recurring revenue to be 10,000 but one customer churn and they were paying 3,000 then that’s your gross churn, 3,000 by 10,000. So don’t blend it, because you’re obviously acquiring new users and if you measure blended, your numbers still look great, right? Because you’re growing a lot, but you’re not paying attention to users that you’re losing. And it’s really important to learn from users you’re losing, which is why we ask you to measure the churn. Paid CAC. This comes in a little later, hopefully almost all of you are acquiring users organically.
I wouldn’t recommend doing any paid acquisition at this stage. But if you do start doing some experiments where you say, “Hey, I’m going to do a little bit of paid marketing or advertising to get a few customers,” then you should measure what was the cost to get that user through paid mechanism. So, which is literally saying, if I spent $10,000 on Facebook or any other channel that you used and you got five customers from that channel, how much did you pay for that? Common mistakes. Two really common mistakes. And this happens again and again, even in spite of like highlighting it, which is ARR literally stands for annual recurring. Recurring is the most important word. Revenue. If you don’t have a recurring business, which is you don’t chat subscription, your customers are not committing to 12 months of payment, you don’t have a recurring revenue business.
So internally, if you start calling it ARR, everyone thinks, “Oh it’s repeat business.” It’s not, because you have to go back and acquire them each month or you have to work with your customers to make sure they pay every month. So the most common mistake is people instead of saying annual run rate, which actually is not very useful even to gauge your business. If you use ARR, be absolutely sure you have recurring gravity. And the second mistake people make is when they would say, “Oh yeah, it’s recurring revenue,” but actually the customer would have committed to only one time payment. Or you may have done a consulting project and it’s not really clear whether they’re going to pay second month or third month, but they would still include it as MRR. It’s not.
Make sure you’re including only what’s truly recurring where the customer has said, “I’m signed up for a 12 month or a six month annual plan and this is what I’m paying.” The third is subscription. This is probably more relevant for consumer businesses. So if you Dollar Shave Club, Blue Apron, Athletech, there are lots of companies, especially subscription as a service is becoming so popular in consumer right now. It’s very similar to SaaS, but it has a slight nuance. So similar to SaaS because you have MRR, which is recurring, right? You may have signed up a Netflix annual subscription. It’s recurring, it’s monthly recurring. The two differences is instead of looking at ARR or any of that, we actually say measure your monthly growth and unit churn, not dollar churn.
Why is that? Because if you’re selling to companies, usually your subscription is more value, right? It’s like they’re paying two thousand five thousand dollars so there if you lose a customer, the impact on revenue is a lot. So you should really look at revenue churn. If you look at Netflix, everybody pays 7 or $10 per month. So it’s about volume of customers, which is why we say pay attention to growth because you need to make sure that the number of users using Netflix continues to increase. And you measure unit churn because if you lost a customer who’s paying $10 it’s $10 but if he lost a thousand customers paying $10 each, that’s significant. Right? So which is why we say measured grows unit churn was as if you’re a SaaS business selling to companies, measure revenue churn.
And paid CAC is very similar like the last time, which is make sure if you’re spending paid marketing, measure the cost associated with acquiring those users. Don’t do it on a blended basis. So when you are at this stage, you will tend to grow in spikes. So this month you may grow 80%, next month you would grow 10%, next month you may grow another 90% right? That’s natural because you’re learning, you’re iterating and you’re trying to figure out what really resonates with the user. So what’s really important is to make sure that since launch, since the month you started acquiring users, you measure compounded monthly growth rate. So which is low, the current month divided by the first month and decrease that growth rate proportionately for the number of months since launch.
What founders often do as mistake is they would just do the average. So they would do 90% this month plus 10% second month plus 80% third month. What happens with averages? It makes your growth look good because you had some spikes. You want to be true to yourself because these won’t be problems when you are two or three people team but hopefully you all will scale and we’ll start hitting 10 people team. And then when you set goals, someone will be very happy that it’s 50% growth rate. But that’s because you’re measuring growth wrong. The next is transactional businesses. This is probably more new, new in the sense that’s happened in the last 8 to 10 years. And again you’ll see that with Stripe, PayPal, Coinbase, Bracks, a lot of Fintech companies especially fall in this category. So what is a transactional company? Which is if you’re in the safe and tech or payment space, you probably process a lot of payments volume, right?
Let’s take Stripe for example, they provide the payments for more startups. And so every startup’s payment volume goes to Stripe. That’s the transactional business. But Stripe collects a fee for the transaction. So if you’re a type of business that processes someone else payments volume, then you should put yourself in the transactional bucket. So the gross transaction volume is the TPV or the Total Payments Volume that flows through the platform. So Stripe had 30 customers at your stage, and all 30 customers were processing say 100 million in TPV, but it all went through Stripe, that’s TPV.
But that’s not revenue. Revenue is what goes to your bank account. So that’s your cut, which is why it’s called net revenue. So the portion of transaction volume that you make, so Stripe would say, “Hey, I charge two and a half percent for the payment volume that flows through my platform,” the two and a half percent is the net revenue that they take. If you’re in the transactional business, it’s very common that you’ll have lots of customers, right? So within a [inaudible] , you could even have 1,000, 2,000 customers. Then the important metric to track is user retention because you want to make sure that six months after they are using you or 12 months after they’ve started using you, hopefully they are still using you. Why? Because you are powering their platform. Unless they’ve gone out of business, there should be no reason they’re not using your platform. How are they doing business?
Like imagine if someone stopped using Stripe. Well, are they out of business or who is processing their payments? So that’s why it’s really important to measure your cohort retention on a monthly basis if you are a transaction business. And paid CAC is very similar to what, in all cases it’s the same thing, just measure from paid channels. But again, as I said, hopefully none of you are doing paid marketing. So what is the common mistake here? Confusing gross transaction volume to net revenue. As I said, if you are processing 100 million in transaction volume, that’s not net revenue. That’s not the cash that hits your bank. Two and a half percent of the a hundred million hits your bank. That’s a much smaller number, so you should really make sure what you call revenue is just transaction volume.
And I’ve often seen founders here sometime come up with like, “Oh, but I process the volume. That’s my revenue.” No excuses. Net revenue is literally the cash you make in the bank and then user retention is a cohort metric. It’s not one number. It’s not like, “Oh, I retain 30% of my users.” That means nothing to us, right? Even for you, it should not. Because if you’re providing payments, you should say, “Well, 40% of my customers have used it consistently for the 12 months since they joined us.” Well, that’s a better metric. Next is marketplace. Again, this is more, it looks like transaction but it’s different. It’s typically used by consumer companies. So Airbnb, Ebay are all good examples of marketplace. What’s a marketplace? You have two sites. So in Airbnb you have hosts and guests.
Guests go to the platform, select a room, book it, the host is happy. That’s a marketplace. So what are the three or four metrics that matter here? GMV, right? So when the guest books the room, the host might say it’s $100 per night. And so the $100 per night, say they stay to two nights, it’s $200. That $200 GMV that Airbnb can record. But that’s not natural venue because Airbnb doesn’t make the full $200. Airbnb probably makes 12% of that, right? So 12% off that $200 is what you’d classify as net revenue. Two other metrics, again, similar to the other morals that we talked about you want to track here is compounded monthly growth rate. If you see this is a more important metric for consumer businesses, right? Because volume of consumers matter and therefore it’s really important to track your monthly growth rate in a compounded way so you can keep yourself honest how you’re growing. And similarly, when it comes to consumer businesses, you should pay attention to user retention, not necessarily dollar retention because the volume of users matter. So here you would say, “What percentage of customers came back to Airbnb or Airbnb retained 6 months or 12 months from now?” Now in Airbnb’s case, how often do people travel? Does anyone want to take a guess? Was that once a year? Very good. So how should they track when should they be happy, when should they be sad? Once a year? Yes.
So if you imagine when Airbnb was going through YC and they could see their cohort repetition once a year, how are you going to know you’re building a good business? They had to wait 12 months. Will you wait 12 months to check whether your customers are coming back? No. So this is the where you have to get creative. And the way to get creative is reflect on your user behavior. So if you’re going to book something, hopefully someone who’s traveling is not booking the day before they’re traveling. They start doing research six months before. And so Airbnb studied that. And so what they would track is if you as users came back to at least search for a city or a booking six months.
In this stage of your startup, what’s most important is to really be truthful and honest about how you want the users to behave and come up with those retention metrics to measure if your business is really healthy. And are you seeing what you want to see from your users? So what’s the common mistake here? This was especially acute for Airbnb because they didn’t pay anything for demand site, right? They had a brilliant value proposition. They were really good designers, very good storytellers. Demand, they didn’t pay anything, but they had to pay to acquire a host because guess what? No one was ready to let their homes to strangers. So they had to work hard, which was they had to put advertising around events. So they had to spend few dollars on acquiring hosts pretty much, very early on.
So the number one mistake founders tend to do here is, you’re acquiring a bunch of users organically and some users through paid and you’ll blend everything. You’ll say, “I acquired a hundred users this month and so my CAC was 12,” but what had happened was if you truly measured who you acquired from paid advertising channel, it could be as high as 70 and so you have to ask yourself, is it sustainable? Are you really seeing the ROI in paid channels? So if you are in an unusual situation like Airbnb there, your business frequency is not very high because people use once a year and you have to pay to acquire hosts. It’s really important to pay attention to where your money is going and whether you’re getting a good ROI from that. Ecommerce.
Ecommerce is literally you have certain goods to sell, you’re selling them online, people are ordering it, Warby Parker, Bonobos, MiMi box, a lot of them. That’s what we would characterize as eCommerce, which is you make the products or source the products, but ultimately it’s your brand and someone’s coming to the brand to purchase it. So here again, it’s a consumer business. You track monthly revenue. Notice there’s no recurring, no subscription, it’s just revenue because people might buy a product that’s mundane and not buy it next month, right? So it’s monthly revenue because it’s consumable business.
Again, very important to track a component monthly growth rate. For eCommerce, even from day one, it’s important to track your gross margin because you either make the good or you’re sourcing it and branding it under your name. So it’s important to understand what it takes, what is your cost to get the good so that you’re making some profit on a per product basis. And it’s more important for eCommerce because it’s not a recurring business, right? So you have to make sure that you’re able to make money on a per transaction basis. So which is why gross margin is important to track. Paid CAC very similar to all the other examples.
Common mistake, gross profit for eCommerce is not accounting for all costs. Now Amazon does a great job of this and people often say, “Oh yeah, they have very thin margins,” but actually it’s net margin of all costs. And so if there’s high volume, net margin times high volume is a pretty good business which you can use to funnel for future investments. So the common mistake we see here is in eCommerce as people would say, “Oh.” Say I bought a clip and we know the cost of the clip is $10 they wouldn’t include shipping costs, they wouldn’t include customer processing costs, they wouldn’t include payment processing costs. All that is important because if you don’t include those costs, you’re probably pricing it wrong. And so it’s so important that you’re pricing it wrong pretty much from first transaction. Advertising, we see far fewer companies in the advertising space these days.
But if you happen to be in that space, then you know the common companies that are analogs for you are Snapchat, Twitter, Reddit, they all have a huge social network that come to their site for different reasons. But the primary monetization model is advertisers advertise there and the companies make money from advertisers. So at this stage, because you probably will never be monetizing if you’re in the advertising business, it’s all about the users. And so when it comes to users, there are really only three things that matter. Daily active, monthly active, percent logged in. So who are the users who use your app daily, monthly active? Who use it monthly? And then percent logged in is actively logged in using a username and password.
The common mistake, and I’ll give many examples of this, not defining what active means. So there was a company, I think three or four years ago, that would put it a daily active user metric. And I remember asking them, “What is active?” And I had some sense of like maybe put somebody who had read, engaged, whatever, and this founder answered, “Well, those are the emails I sent.” That’s not active. Active, again, it goes back to the Airbnb example. You should define what you want your users to behave like when using your app. So if you’re building a news app, does it mean reading counts as active? Does it mean commenting counts as active?
You should define that really well. And so if you don’t define that, you could be building something that has no stickiness and probably you’re going to have users that are winding down pretty quickly and it’s not worth it, right? So make sure you really define what active is and hold yourselves to that metric. The other one is hardware. It’s again, very similar to eCommerce because at the end of the day they’re selling a device. So if your Fitbit, GoPro, [inaudible 00:22:41], we’d say you’re in the hardware bucket. As you can see, it’s very similar to eCommerce where you look at monthly revenue, compounded monthly growth rate. You look at gross margin really carefully.
Hopefully you’re making profit from day one and then paid gap. Okay, so those are the nine business models. And one last thing I’d leave you with before I open it up to questions is common mistakes. So common mistakes is, you’ve heard this and you’ve probably read this in so many blogs, charge that look up into the writer. Brilliant. Well, but accumulative charts are always up and to the right. Do not have any com, there is no rational in the world to have accumulative charts. So I don’t know a single company at scale that shows accumulative charts. So do not take accumulative chart.
Second thing I’ve seen is not labeling Y axis. As I said, you’re going to scale even if you hit five or seven users. If you don’t know what the Y axis label is, and if the charts look like straight vertical bars, it means nothing. Third is changing Y axis scale. This is something I never understood, but quite a few of them do it, which is X axis starts at zero and Y axis starts at safety. Those things don’t really show how well you’re growing. Show your problems, by the way, no YC startup had a chart straight up and to the right. No one did and the most successful companies didn’t either. So I think the most important thing is to really be honest, measure and fix things, right? It’s okay to go down sometimes. And also usually we say don’t show only percentage charts.
It’s very important that whatever you’re measuring, whether it’s gross revenue churn, monthly growth, be clear about the absolute number and the percentage relative to the absolute number. We also had done detailed posts on metrics while I was at a16z. So I’ve included two links there if anyone wants to look at it. But at your stage, only three or four metrics matter. If anything you took from here, hopefully you fit into one of these nine business models. You can start with two or three of these metrics for each business model and that itself will be a great headstart for all of you. Thank you.
I have two quick questions. On the enterprise side, but you almost equate the bookings to accrual accounting. Is that a one to one matching right there?
Yeah. So the question is can we map bookings to accrual accounting, right? Yeah. I think that’s not, I mean, the purpose of tracking bookings was as revenue is not from an accounting standpoint, it’s from how do you gauge your business, meaning bookings as future revenue, right? So all you’re saying is, “I have signed a contract for $a hundred thousand. I am not ready to launch for them until two months from now.” But you know that it’s future bookings because it’s all signed and the customer has committed to working with you. So you say it’s a hundred K and then, but you’re not starting to deliver service until two months from now, which means you can’t recognize revenue. So it’s only a measure to manage your business internally. So you know that this is coming in the pipeline. This is where my team’s resources is going to go. But if I want to hit the goal of 200K by the end of the year, how should I, maybe you need to go find one more customer or accelerate launching your service, which is most likely the case in the phase that you are in, which is why am I waiting two months to launch this? What can I do to launch it tomorrow?
And the second quick question was for community business, we can pick from a couple of different things within these nine verticals and business model. Is there something that you are recommending for a community business to track early on before you decide business model?
I think that the question was really what should we track for community business? I put community business whether your business model is advertising or not in that category. So really what he should be tracking is define who your active user is and what you want that user to do. So daily active users and monthly active and track weekly how they’re doing.
yeah. You know, most businesses evolve to different business models over time., right? What’s most important is what is your short-term focus? So if you’re short-term is like for the next year or two, which is about finding product market fit in your transactional business, then that’s the area you want to start with.
Yes, absolutely. So the question was what should we talk to for investors. At this stage, honestly, the bet is only on two or three things, which is you as the founder. What unique insight do you have that you’re building this and can this be a big company? That’s it. And so don’t overcomplicate this.
Do you have any suggestion
Yes. Yeah. So great question. Again, as I said, it’s very easy to get caught up in what do investors want to see. But the most important thing is you know more about the business than anyone. And so it’s about what should you track to make sure your business is healthy. So the two things as founders you should track very early on is burn, right? Because you’re spending cash. So you should always know, don’t run out of money. So there’s no excuse for running out of money. So make sure you do your P&L quarterly so that you’re tracking your burn and you know that there’s enough money in the bank to at least operate the company for another six to nine months. Two is related to expenses. It’s the gross margin, especially if you’re in business categories that are eCommerce or consumer related. Then they do pay attention. It’s important for founders to pay attention to gross margin which is on a per transaction basis. Am I making money or not? Because if net you’re not making money then it’s, do you have line of sight when you would make money because if you don’t then you should really question is that a business to be built? But I’d suspect it’s quite early right now, which is why I say focus on getting the users and how do you monetize? That’s the first step.
For businesses has a model kind of like take an ancestry.com where people sign up. They then report and they may want to come back and refer to it later but there’s new addition revenue for what they originally purchased. Would you do that sort of eCommerce business? Would you also monitor their-
Yeah, I think you should do both. But I would record revenue as just the revenue for that month because it was purchased only for that month. And then you just figure out like how Airbnb did, which is what’s the behavior you expect? Do you expect them to come back once a year? If so, are they and what percentage of them are coming back? Yeah, I’ll go to this site.
Yeah. So great question. And this is why I brought up the Airbnb host example, right? The only thing that I would say that’s probably most important is to measure the paid acquisition cost of bringing that supplier in. Because in most marketplaces it’s actually really hard to get the demand side, which is the users, right? And if you get enough business for the supply, the supply will stay. And especially at this point, again, if you are three years out and you have scaled and you at like 20, 30 million GMV, the answer to your question is different. At this point it’s about how do I get both sides organically as far as possible. If you have to spend on the supply side for a little bit, okay, but try experiments every now and then. But when you’re trying experiment, the most important thing to measure is what’s your paid acquisition cost because hopefully it’s not so much that your born goes out of back so that those are the two that really matter.
Wait, you said about enterprise business that you know on day one we won’t be selling to Facebooks and Googles. So what’s your recommendation like break down the business model? Just go down to smaller teams of individuals in the enterprise.
In given an enterprise businesses, you can’t really sell to Facebook from day one. How should we focus about the go to market? Should we just go to small teams? You almost always start with small teams or a pilot, right? So when companies do focus on enterprise, you usually start with a pilot. But the pilot could be something like it’s a six month pilot or it could be a three month pilot. In those cases, you don’t report them as bookings or revenue. You just say it’s a pilot. At the end of the pilot, we visit whether we can get a contract. And the phase that you’re in, usually you will have only one or two customers over the course of 12 to 24 months. That’s okay because the ACV, if it works out, the annual contract would be really huge and that’s expected. One more.
Alright. I know that’s probably my answers to this question but how do you determine what a good customer acquisition cost is?
Yeah, it’s really hard. The good acquisition customer cost is zero. So let me just say that. That’s the best customer acquisition cost, especially at this stage, because if you have built something that people really want, you will get it without spending much, right? So at this stage it’s only zero. As you scale, there’s obviously room. It’s not necessarily, the correct answer is not an absolute dollar number. The way you look at it is you compare your acquisition costs to the lifetime value of the customer. So if you had the customer for one year, in Airbnb’s case, for example, if Airbnb says, “Yeah, you know, this customer booked twice with me, and if they do that every year, I make $100 from them,” then you say, “I can, you know, you should have a five X gap.” So you’re willing to spend $20 to acquire, but at this stage the answer is zero.