Y Combinator Startup Podcast - #54 - Growth Office Hours: Round 2

Episode Date: December 15, 2017

Anu Hariharan and Gustaf Alstromer are partners at YC. This episode is a follow-up to Anu's Growth Guide and the first Growth Office Hours.The YC podcast is hosted by Craig Cannon.

Hey, this is Craig Cannon and you're listening to Y Combinators podcast. Today's episode is with Anu Hari Huron and Gustav Alstramur. Anu and Gustav are both partners here at YC, and they came in for round two of growth office hours where they answered questions from you guys on Twitter. All right, here we go. So let's start with when should you build a growth team? What do you think? I think, you know, most companies think about building a growth team when they have strong retention.

that's, you know, we touched upon this in the last office hours as well. But when exactly is that, right? You know, for many companies, it's roughly when you have maybe 15 to 20 people in your team, at least two product managers. And if you've figured out that you have strong product market fit with retention, then the third PM that you bring on the team might actually be a growth PM. And so that may be the time when you start investing in a growth team. But, you know, it's never too late to invest in a growth team, but too often the mistake I've

seen most companies in Silicon Valley do is they actually wait too long. But it's not, you know, it's not like, it's not like the consequences are then negative as long as you realize that the team is actually helping and you have formed the team and you know, put them in place in the right direction. The companies that have actually done this early have benefited a lot from it. So Slack, to give you an example, was actually one of the very very. very few companies that hired a growth PM 14 months after launch. And I think it was roughly 1 million DAUs, so fairly early in its evolution.

But it really helped, you know, having a growth team in place helped accelerate the entire company towards scaling it from like 1 million to 5 million due AUs in less than two years, which is no easy feat. So the sooner you invest, the more benefits you get. But the most important thing is you should know when you're ready. and for that you need to look at retention. Okay. Would you agree with that? Yeah, I agree. I think the biggest mistakes people make is that they don't look at retention and they don't really know when they have product market fit.

And they fundamentally haven't built something that people keep using on and on again. And that's not the right time to build the growth team. You still have to figure out why is the product not good enough? Why is retention so low? Why is it decline to zero effectively? Once you figure that out, growth team can help you with many things, but there are many specific funnels, the drives towards your product value,

that you can accelerate. And I mean, I've seen pretty much every single step you can make a massive increase on if you just have this laser-focused data growth team can have. Okay. And do you have benchmark retention numbers that you advise people to look for at the, yeah? Yeah, I mean, I think that's another mistake companies do,

which is they don't really benchmark retention to the peers in their space are vertical. and it's extremely important to do that. So to give you some examples, you know, for example, you know, even if you are building a social network, right,

you should have long-term retention of 50% or more. Because if you're building a messenger, someone should be checking in daily. If you're something like a Facebook, people still, 50% of them still log in daily. But if you have, if you're building a social app

that has less than 20% retention in month 6 or month 12, like that's worrisome, right? So you should really pay attention to what your sort of space or vertical is in. Now, if you're an on-demand delivery company, the median retention, you know, long term is usually 20 or 30%. So they don't have to have 50 or 60%. But for them, 20 to 30% means it's probably good retention. The general rule of thumb we often tell companies is if you have anything less than 20%,

like you've got to start paying attention, you know, no matter what space you're in. So there you have some sort of a benchmark, depending on wherever you are, you can sort of pay attention to, like, is this really weak retention or not? But beyond that, you have to start paying attention to what space you're in. And sometimes the periods also differ. So for on-demand delivery companies, we actually look at it depending on food or right-sharing versus grocery. You can actually do it weekly versus monthly. You eat food weekly.

You tend to order weekly. Right-sharing, you take trips two or three times a week. so weekly makes more sense. Something like travel like Airbnb, it's more yearly. You don't travel every month, right? So defining the period and then benchmarking what your long-term retention is versus your peers become extremely important. So that's tricky.

So say you're Airbnb and you're only a year old. Do you, what are you looking for in retention if it is actually an annual thing? Well, that's a classical problem when companies go and raise money is that, oh, my retention curves or my retention data sets too small. that's why when you start a company you should start tracking those key metrics that matter for your business, if it's bookings, if it's deliveries,

if it's orders, or whatever it might be, just make sure you track them from day one because they will matter and someone's going to want to look at them when you're raising money. If you, let's say I've been at it for a year, like say, every's first year and you look at the retention cycle as yearly,

well, there's a lot of predictions probably will predict like the next booking and you can start looking at them. At some point you're going to be like, yeah, this seems to be a strong enough prediction, let's say coming back and searching with dates. That's a strong enough prediction that I probably will book if I find something that's good.

Okay, cool. And on that point, I would add, like return visitors is a way to sort of engage that versus booking. One of the things you still have to keep in mind, you know, and I was there even in the early days of Airbnb when they were raising. Most investors could not comprehend that because they didn't know is this really sticky or not.

One of the reasons why it took, you know, Airbnb may be harder in the earlier years to race was exactly that because it was travel, velocity of views is really low, and there was not enough predictability on retention. So it's a great case study for founders out there if you're wondering, you know, should I keep at it or not? Find other ways to measure customer engagement. And you can teach the investors on what those metrics should be and how you are building confidence that the retention will look good. So I also want to clarify that the term retention might, to some people sound like,

oh, this is just the growth team's kind of like their own metric that we don't have, I don't care about growth, I have to worry about retention. Well, think about it. What is retention? It's like a representation of the value that someone gets out of your product. And no matter if you care about the word retention and not you care about growth theme, you want to build a product that people get a lot of value from. And it turns out their retention is just a really, really good way to measure the value you get out of that product.

So it actually applies to everybody. No matter if you believe in what growth teams does, if you believe in the term retention, it just applies to everyone. It's just not a choice really. Just making a better product, basically. And so at the point that you've decided that you're going to build that a growth team, what does that first hire look like?

What do you look for in that person? So I, well, there's one thing. There's the personality of the person you want to hire, and then there's the discipline that are all you're looking for. Okay. Generally for personality, I look for people that are curious. So a good example of this is like, what's the first website you look at when you come in the morning? Well, if that's the dashboard, that's the experiment analysis, you're a curious person.

You're curious to see what the results of the thing you shipped yesterday is or what the metrics are doing or try to kind of tie in the numbers with the thing that you're building. So that curiosity is super super critical. The second thing is just someone who has a low ego. like you're going to have to throw away a lot of code. You have to see a lot of results that basically might be contradicted to what you believe. And you just have to like deal with it because like you don't actually know. You don't actually know what's going to work. And you have to like be fine with trying a lot of things.

Whether that's an engineer, whether a designer or a product manager, you just have to try a lot of things. And then someone who's who can quickly get to the point of like what's an MVP or what I'm trying to build. What is the smallest possible test I can run to test this hypothesis that I have? and it usually doesn't mean you have to spend a month building something. It usually is a lot less. Now, in terms of the disciplines, the most important person is an engineer. So if you're just a product manager or just a designer,

you can't really do anything. If you're just an engineer, you often can ship something. So usually it starts with engineering. The critical roles after that is someone who's data savvy doesn't have to be the product manager. You can start with the engineer itself, can start with an analyst or data scientists or a product manager. and then you want

Basically you basically want these things to come in hands So start an engineer, get someone in an understanding of data Then maybe you add a designer Then another engineer eventually a product manager Kind of comes that way But very often product managers plays this early role of leading this team Which is typically why product managers are often

In the very, very first incarnation of these growth teams Yeah and that's something we heard When we were talking to a lot of the growth teams as well One additional nuance which Gustav did add but I'll mention it, is Gustav himself is a former founder. And we actually saw that, you know, at least 40% of the growth teams that were led in Silicon Valley were former founders. So we asked some of the CEOs why that was the case. And it goes to the point that Gustav mentioned earlier, which is you have to be self-initiated

and so comfortable with throwing away a lot of court because a lot of it is experimentation. So you need someone with a lot of grit who can withstand a lot of failure and would still keep pushing. to figure out the one-third of those experiments that's going to make a meaningful difference. And where do you find that characteristic, usually in founders? It takes a long time to persevere. So I think it's no coincidence that you see that. It's not a must-have. We've seen growth teams that have been very successful without founders too,

but it is an interesting observation that a good chunk of the growth teams are actually led by former founders. And we've also heard that from CEOs saying they explicitly try to hire a former founder in that growth PM role if they can because they are looking for that experimentation mindset. And the reason I got into growth in the very early days was that I was building a product that wasn't growing. So I had a problem. And let's say you worked in a very big company that never had that problem. Growth is just free. You just always have growth. Well, then you're not going to start thinking about what matters to growth. You're not going to look at these forms. You're not going to look at this specific funnel and try to care about what matters because growth is always there. But for a founder,

everything matters, like literally every single step matters when you're starting from scratch. So I think that's probably the reason is why this works really well for people who found their background. And how do you found these people? I mean, you obviously can read their resume and see founder, but what were you guys advised when you were like surveying other companies? Yeah. So when we talk to a lot of companies, you know, I think the first step they try to do is see

internally whether there is one PM with that sort of mindset, especially the data-driven curiosity, using data to help answer, versus, you know, just going with product intuition. Because I think it's very important for this growth PM to look at a lot of data. In fact, you know, George Lee, who was the head of growth at Instagram, puts this really well. He says, the scariest day is when your metrics go down. But the second scariest day is when your metrics go up and you don't know why. So the growth PM really has to have that mindset of why. What is really driving each metric that day and truly understanding the behavior behind it?

You know, they have to have the grasp for that. So the reason they look at internal PM first is simply for one reason, which is social capital, you know, meaning, you know, the growth team has the onus of, you know, working with all other teams in a more collaborative way. Because if they were just focused on what they were doing, it probably wouldn't work. You're running a lot of experiments. You may need to make tweaks to the product. You may also have to make design changes. And when you're often establishing a team early on, you don't have all the resources. So you've got to work with the product team.

You've got to work with the engineering team. So having a PM who already has that internal social capital and can debate ideas on experiments and convince people on why that's required is an easier move. But not all companies are able to do that. So sometimes you have to hire externally because maybe the 2 PMs or the 3.000, PPMs you have on your team don't have that growth mindset. You know, and it has worked in other cases too. But I think the most important lever you need there to make sure the new person that comes

in is successful is that endorsement from the senior leadership, especially the CEO, that it's an important function and you want everyone to work together. And so to dig into that a little bit more specifically, were you guys, you know, trialing people at Airbnb or if it's just your first growth person, are you having like multiple candidates do like a month at a time competing with each other? How do you actually pick the person? So I'll give you a book recommendation. There's a book called Work Rules by Las Laboc, which is probably the best book I've read about hiring.

And there's two things they talk about in that book. Basically, it's trying to answer the question, what makes a great hire. Two things that comes out of that is one is cognitive ability. like problem solving skills effectively. And the other one is like, can you do the work? So when we did hiring Airbnb, the second was very, very critical. Basically, can you do the work? So we would put them, like for PM, we were just like,

here's a growth problem. Most likely something that we actually are facing right now or something that we faced like a couple months ago where we know a little bit more about. We basically asked them to give us your best recommendation, like how should we solve this? And the output of that is maybe you spend 24 hours making a presentation, and you spend an hour presenting it for the panel.

and basically that helps us determine can you do the job in a much better way than you would do an interview. Interview you ask hypothetical questions, but at the same of a real use case, you spend an hour actually talking through things, you can ask a ton of detailed follow-up questions on very specific things. So that's basically what we did.

Once you've hired someone, you've hired someone. I'm not sure if I believe in like three months trial because you have to give up a job in order to get hired somewhere. So there are some downsides with leaving your previous jobs. In terms of internally, so the way it worked at Airbnb is that engineer would join Airbnb and different teams would pitch that engineer. And we would look for those characteristics that I mentioned earlier for the kind of people that would be good for growth.

And we basically would spend time with them and kind of figure out if they fit the kind of person that we were looking for and not everyone did. And that's totally fine. There are lots of other teams you can work on. But some people were like, I want to do growth from day. They just came to us. I want to have been the growth team. Yeah.

And then that's just pretty awesome. And that was true for you as well, Anu? Yeah, I think that, you know, different companies obviously have different models. I have seen at least two companies that actually have the model of come work with us for three days, not a month. But, you know, even if you get someone to work with you for two days, you have a, you get a lot more colored into their personality, how they think, how they collaborate with other team members and how much of the job do they really understand, right? So those are elements you can get a sense of. and some really young startups do that. They, in fact, offer it as a paid vacation day

or ask the interviewees to negotiate that, but they do it obviously with a select field, so you can't do that with everyone. That was an interesting approach I saw with a couple of them. In other cases, you could probably, you know, today there are so many growth teams out there, you can actually hire someone from a growth team. Like, you know, Facebook has more than 2,000 people

and a lot of Facebook growth folks are now at other companies. And similarly, you know, you can hire now from lots of other teams, right? Airbnb's growth team is 100 people big. Uber's growth team is 500 people big. So I think growth teams are no longer a novel concept. And so founders have said that they find it fairly easy versus five years ago to sort of tap into that, you know, growth PM recruiting pool. And is it fairly easy to suss out the people who are very effective on those growth teams? So you're hiring someone from Facebook and you're like, we're growing, pick a product like 10% month over month. How do you know what's,

that person that you're hiring, that's affecting that change? Yeah, no, that's hard, right? It's not easy to do. But I think that's where Gustav's example of presenting your problem during the interview and seeing how they think through that and react gives you a better pulse for, is this person really wired for what we are trying to build? Because, you know, many people assume that if it's from Facebook, they'll know everything. They'll know all the tools.

But it was a, it's a social network at the end of the day. And the way you grow a social network is very different from how you grow. e-commerce or how you grow something like travel, which has extremely low velocity of use. So I think you have to always test the person in the realm of your context to test out two things. One, do they have the depth of understanding that you need about the problems you're solving? And are they able to bring their own creativity versus just saying, at Facebook, this was the experiment with Ryan and we need to do it, you know, because it's not Apple's stuff. It's absolutely the approach.

That's like the key thing here. So like if you work at Facebook, then you will be like dreaming. beaming DAU, MAU, and like those kind of metrics, which actually didn't apply to Airbnb at all. Those are not the kind of metrics we cared about at all. So if that's, you've got answers, like, oh, I'm sure looking at these metrics when I'm evaluating a party of MB, then you weren't thinking hard enough about the type of product that Airbnb was. So I think the signal you get from having someone spend an hour presenting a solution, we don't expect you to know the specific channel tactics or any of those things.

Like, that would be great if you do, but you don't, that's not really what we're looking for. We're looking for the creativity and the ability to solve the problem and come up with like, how would I, like a common question in that challenge would be what are the metrics that you would use to evaluate whether you have solved this or not. And that's kind of like, it's not like you just have to think and use the product and you can come up with what those metrics are. Gotcha. Okay. And so now let's assume we have a growth team going. When should we start thinking about paid marketing? Yeah.

I think just because you have a growth team doesn't mean you should start doing paid marketing. lot of companies actually didn't. You know, the growth team can actually really spend a lot more time on the one or two organic channels that, you know, the product has been doing very well on, including referrals. You know, Gustav, I'm sure, can talk a lot about referrals from Airbnb. But I think the big mistake I see companies doing, the moment they hire a growth team, they feel like they have to give them a marketing budget, and that, you know, that suddenly the company has to go from organic to paid. And that's the biggest mistake I've seen. Because, you know, paid marketing will give you some results, you know, for sure.

But you wouldn't, you won't know the retention of those users for a while. And so how do you know whether all the experiments you're doing and the acquisition of these customers is good or not long term? And second, if you're a startup, the longer you can wait to use paid marketing, you should. To the extent you can actually get a lot of it from word of mouth or, you know, be it or through even SEO or referrals. You should tap into those first before you turn into paid marketing. You can always experiment a little bit with paid marketing gear in there just to get a sense for which channels you might tap anew. But at least in the second and third year of your startup,

that should not be your main channels. I would say the amazing thing about a paid marketing is that it's the tactic you can deploy with zero time. Basically, start tomorrow and you can buy lots of stuff. That's great. Most channels don't have that opportunity. The biggest problem, though, is that if you aren't making the money back to your spending, you don't have anything.

Like, literally, you don't have anything. So let's say I bought 1,000 users, and I'm not charging enough to make the money back to spend on them. Well, I really haven't learned much because I can't buy another 1,000 users, because eventually that strategy will fail. So the only only way that the paid marketing works, if you have high LTV, high enough that you can recoup the cost of the marketing that you spend, that's the only way it really works and you kind of need to be sure that those cohorts and the retention of those paid

marketing users are actually worth as much as the previous ones. So once you get the answer to that, then you have a lot of opportunity. The hard thing at that point becomes around measurement. So the more different marketing channels you have, you have to be pretty good at measurement. I mean, it's not like it's rock and science, but you have to do it and people fail often in the measurement side in the early days. If you have a free product, you won't really learn anything from your marketing. You just shouldn't do it at all. Like you might do it in the product kind of customer development phase. You just get a few users, but that's when you're talking about less than 1,000 users.

But if it's a free product, marketing won't really help you. And so aside from, you know, jumping into paid marketing without knowing these stats, what are the other big mistakes people make? I think, let's see, I see some mistakes around the word incrementality. So basically people think that, buying branded traffic on Google with my brand is incremental. No, it's not incremental. It's usually not even marketing.

It's a result of your product growing through a different mean. Let's say, I told you about Airbnb and you went to Google. You typed in Airbnb and clicked on the brand ad. That's not marketing. That's just actually a different channel of the marketing. The marketing piece here is when you type in apartment rentals in San Francisco or vacation rentals in Miami. That's actually the real non-brand marketing.

So that's the incremental marketing that you were getting that otherwise wouldn't be getting. And that's why it's key to think about experimentation. This is true for say remarketing. So remarketing, there's a lot, like, every single click you get from remarketing isn't incremental either. So you have to think about a way, how can I actually measure myself to think,

to know what, the clicks that I'm paying for are actually people that wouldn't have otherwise gotten to my product and started using the product. So the word incrementality I think is just something that people can tend to forget. And there's basically the answer to most of those things is experimentation. Yeah, and I think the second one I would say is coming up with metrics, the right set of metrics,

and holding yourself accountable to even measuring the results of the experiment is another area that I've seen mistakes happen. Too often people are just focused on acquisition of customers and not the LTV or the retention, especially after the experiments. And if you don't measure that, you actually don't know whether that experiment was good or not before you sort of make it mainstream. But I've seen time and time again companies do that mistake. where they just look at the growth and they forget about measuring the attention of those users,

even two weeks down the line or four weeks down the line. If you can't tag the users that you bought through paid marketing and you know downstream what they do compared to the people that don't come in, you're lost. The numbers might look good, but essentially you're lying to yourself, right? Like you have no idea. Very often specific, I mean, specifically display channels will look radically different than any other users you acquire from any other channels. People are like, oh, I spend a ton of money on display marketing, and now they don't perform

their way that every other user do. It's like, duh, of course they won't. But search marketing might be a little bit different, but display, they will be radically different. Yeah. The other mistake I've seen companies do is when they establish the growth team and even do paid marketing, they will try all channels, you know, fragmenting the budgets across four or five different channels.

And if you ask companies that have done it for real. well, they would say for a long time, two channels gave 70% of their user traffic. So, you know, focus is extremely important. It's easy to like go and spend across six different channels, especially if you're an e-commerce company, you have influencers, you have YouTube, your Instagram, you have Facebook, you have Pinterest, and you think like all of this looks cool. But it's more important to really identify those one or two channels that are scalable and really focus on optimizing those versus doing six things that early. Yeah. I mean, that's exactly what we do I see like I've tried almost every channel but eventually you realize

that you're way better than your competitors at certain channels and you just go for those because that's where the numbers are coming from like you know like we could hang out on snapchat all day and like maybe that would work if we were the best in the game but if you're just average it's so easy to just fall below also turns out that vast majority of the growth of online marketing happens on Facebook and Google so it like we are going towards consolidation here that you most likely, if you become a billion-dollar company, we'll figure out one of those two channels. So let's shift from kind of the founder perspective

to the employee perspective. So Ken Ruff asked a question, if you're working at a company that is not very data-driven, how might you go about trying to convince your leadership team to prioritize data and potentially start a growth team? What do you think? Yeah, so I actually ran into this quite a lot when we were interviewing a lot of growth teams.

Honestly, I don't think I have one magical answer for this. I think in talking to a few teams that were able to influence this, what they did very well was, you know, a communicating, even if they need be over communicating. So this one particular team gave me an excellent template of an email that they sent to the whole company. And I think they very strategically did it at Saturday at 7 a.m., even though the email was done by Friday. And it would just highlight stats, but in the context of what drove the launches that week or experiments that ran that week. And they would have fun anecdotes on, you know, the outcome and their rationale on the outcome. And one of this growth team leads told me that sending it Saturday 7 a.m.

With these nuggets and data really caught the attention of the CEO. And so sometimes he would engage by forwarding it to the respective folks saying, why did this come down? You know, he just wanted to know, right? Because this was the one snapshot that he was getting about how the week actually went. So that's one way I have seen, you know, some teams have gotten to influence at least the rest of the company.

And it actually changed the culture of other teams in that company that everyone started sending sort of a data report. And so it doesn't mean your decision-making moves towards only data. So I think that's the constant tension. The growth team uses data, and then you can have that. the product team thinking, well, why is everything based on data, shouldn't it be based on intuition?

Yeah. And I think there's always a balance. And especially early in the startup, you're still building out a large portion of your product roadmap. You need to have gut and intuition about what the customer wants. Not everything is from the data. So it's right to have some sort of balance. But to ignore data completely is the other side of the challenge.

And so in some cases, it has worked. Honestly, in two or three cases where I spend a lot of time with the growth leads, it didn't work. The CEO's just never engaged on it. And if you don't, if you can't align with the leadership team, you know, I have personally seen those growth leads leave. And it doesn't mean it's a make or a break. You know, the company can still figure out other ways to do it. But there's no one magical solution. The leadership and especially the CEO has to be inclined to buy into it. I couldn't agree more. So even if you believe, let's say your company who doesn't pay a lot of attention to data. There's always

data. It's just not surface. Like the data is always there. You might not be tracking it, but it's there somewhere. So the first thing I would do at that point is just start making sure you track the data. Make sure it go somewhere that people can look at. Maybe it goes in an email, maybe it goes in a database, maybe it goes in a dashboard. Just somewhere people can look at it.

The second thing is like, I think I'm assuming that this comes from a company that's doing okay. Like if you basically are doing okay and you're not looking at data, then you're confusing correlation and causation. You're like, oh, I did all these things and the company's doing great. The thing is you don't really know and you really, really don't know and people are, there's so many of these confusions that people have around what cost the company to be successful.

And the only really way to do this is to look at data. So I think to Ken and this question, I would basically get the data, find a good place to send out the data. Maybe that's email, maybe that's a dashboard in the wall, maybe that's something else. Just make sure the data gets out there and then find a way to prove out what is actually driving your growth. Like, what is the number one driver? It might not be the latest launch you did. It might be the fact that Google AdWords is working really well or the funnel is working really well,

or you've got a better payment mechanism. It might not just be the latest launch. So I think at that point, just get the data and then find a way to service the data in the right way. Yeah. One other nuance that'll add to that. you know, often the problem is you use data and you expose problems. No one likes problems. People are looking for solutions. So one tactical way, especially if you're trying, finding it very hard

within a company, to get someone to listen is show the solution with data. I mean, you know, there's a famous story about how, you know, Uber always said they would never do cash, but they went to India where there are no credit cards. So how do you convince the team that did not want to do cash to actually allow cash on delivery. Or, you know, to do a different form of integration. Run an experiment and prove the adoption is really 10 times better than the alternate. That's another way. And then the person really has to explain why they're still going to turn down your idea.

Right. So make it, you know, go with solutions if you're finding it harder than just exposing problems. And ultimately, eventually, eventually you're going to get to a point where people can run these examples. these experiments yourself. And until you have the permission to do that, I would just do it anyway. It's like run of one of these small experiments in one country. Like at Facebook,

they were running tons of experiments in specific countries all the time to try new things. I would just like run an experiment and see if you can prove out something. Yeah, usually, I mean, at a company this size, you would probably already be able to find people that you're working with

who want to start doing this stuff. So you just dedicate a couple hours to an experiment. You're good from there. Okay, so Kate Nolan asked, a question for you in particular, Gustav. And what does a product lead for growth actually do? Which was your previous title at Airbnb?

And why did the team decide that it needed to exist? Oh, so my role at Airbnb before I left was managing the product managers on the growth team. That's basically the product lead role. We had a number of different teams. Some of them were more, required more domain expertise than others. So there were some teams that are on paid marketing or on SEO, maybe referrals required some specific experience in those areas. But in general, they were growth PMs that could move across different products

in different areas pretty fast. So it wasn't like there was super deep domain expertise. But effectively, a product lead for growth that's just a product manager with a little bit more responsibility. Fair enough. Okay, that's an easy one. We do have more Airbnb questions, but we can kind of expand this to cover the growth post, Anu.

So Tim Knight, the following question, asked when Airbnb moved internationally, how do the strategy change from what you're doing in the U.S.? So this is almost before my time. Now, Airbnb is a little bit different than most companies. Airbnb has a global network effect. Most companies do not have that, which means most of the countries, Airbnb kind of entered. When they set the first foot on the airport in those countries, they already had hosts and they already had bookings. Very rarely they had nothing. most companies do not have that effect.

So most companies have to do something to grow somewhere. You have to make sure your site is available in that language. You have to maybe if you're a marketplace, launch in a different way or add payment mechanisms. Airbnb didn't really have that problem. We had several different organizations driving international growth at Airbnb. On the growth team, I can speak to with the things that we found the most important. Number one, translation.

So translation is super, super critical. There was a guy called Jason Katz-Brown, who, before I even joined, had launched Airbnb in 24 language or something like that. He built it single-handedly himself. There's some stuff you can look at online around this, how he did that.

But that was absolutely critical. It's critical for conversion, because if you don't understand the site, you won't know really how to use it. It's critical for SEO. If you aren't translated, then people type in something on Google in different language,

they won't find Airbnb because you won't rank. So translation was absolutely critical. Then in general, when we thought internationally, we had a strategy that the first layer we said what we called product gaps. Is the product broken in a certain country? Like something about the experience broken.

Very obvious things would be, well, China, South Korea, and Russia would be three countries where a lot of things were often broken if you didn't do anything specific to them. So Facebook doesn't work, Google doesn't work, Google Play might not work. There's a bunch of these things you kind of have to change. Most other countries,

there are fewer, like the infrastructure are the same. So people use Facebook, Google, a lot of the other infrastructure that we would use in U.S. is available and accessible in these countries. So you just have to do some minor optimizations around site speed and translation and those things. So that's the first thing, the product gaps. Fixing that and fixing translation solves most of it. I am not 100% believer that you have to do a ton of work to make the culture of a certain country accept a product.

So if you fix the product, fix translation, not saying that, oh, this concept doesn't work in my country. For my experience, most products don't have that problem. Most things that grow big in most countries will eventually grow big in all countries. Like there are very rarely cases where like, oh, this country don't have to change the product dramatically. It's like a cultural difference and it will never click. I just don't really, I just haven't seen too many examples. However, that seems to be the number one thing that people say, oh, Japan is so different. I've got to change the entire product in Japan.

Usually it's not the case. just usually not the case. And usually use the research will reveal these elements. I think a lot of time, if you rely just on data to form your opinions, that's when you can run into this black hole. So, you know, Instagram, for example, when they launched in the Middle East, this product, which is, you know, heavily used by women in the U.S. and another developed countries, all of a sudden saw a huge spike in male profile users

in Middle East, and they were wondering why they're not able to recognize. the women, but they launched a user research team that traveled to Middle East who then found out that a lot of these users were women, but they were afraid of harassment. And so their pictures on Instagram or the profile pictures were all men. So, you know, you have to be on the ground. You know, no amount of data would have told you because data would show that 90% were male. Right. But you have to be on the ground doing focus groups and user research to understand, you know, why is that the case? And can the product make small,

tweaks or, you know, small changes in their profile appearance for those, you know, to respect the elements of that culture. Were there any other radically different techniques people used internationally? I mean, I think I've also heard, you know, I've heard about Japan consistently, but I think it's more, it's not like the product changes completely, but what people are comfortable sharing might be different. Like, people are probably not as comfortable sharing pictures, you know, publicly in Japan. So if you were using that as a tactic, like Facebook in the early days was using tactic as, you know, photo tags as a way for you to come back and sign into Facebook.

Like, it wouldn't have worked in Japan. Right. So those are the elements you have to really pay attention to. It's not like a radical, dramatic difference. You don't have to change your product fundamentally. Yeah. I guess like QR codes, right, in Asia took off. Yeah.

Yeah. I think that... I think it will, though. I think that's a great example where like we just haven't seen that yet, but we're all going to use in QR codes. You know, often people say that. like to me like oh US will never accept QR codes because we don't like QR codes I think it's not the case I mean the two simple reasons US does not use QR code is iPhone and Facebook because iPhone did not come and come with an inbuilt QR code reader it didn't make it it made it very hard for

anybody who had a QR code and Facebook which is the app for the largest usage did not have offer a QR code in China you had the phones with inbuilt QR code readers and Vech app for it and so if if you make it more mainstream yeah And it improved the ease of use and convenience for the user. Why not? So, yeah, I agree with all that. Here's a common mistake that people make. They're like, oh, I want to grow internationally.

I did that. It's not working. There must be something with that country that's not working. Oh, people don't like to use Facebook a sign-up option. That must be the main reason why it's not working in this country. People don't use credit cards. People don't like to use phone numbers for sign up. Well, I got all of those things over the years,

and we ran experiments trying all of those things, and they are marginal differences. Like basically you change one payment method, you get like a 5% or 10% difference, but it's not like the majority. You change Facebook to not Facebook, and you get a small, small change, but it's not dramatic.

We even launched phone number signups. That's also a pretty marginal change if you already have a bunch of signups. So these easy cultural solutions for why my product isn't growing in those countries rarely is the case. Now, we should try to figure out how does it grow in the first country?

Like if it's growing the first country thanks to Google AdWords or SEO, and it's not growing the second country. Well, maybe I don't have any SEO in the second country. Maybe I don't have any AdWords actually in the second country. But my marketing isn't doing anything there. Like that might be more... So again, it's like returning to the data in knowing why you're growing. Yes, yes. And it's not like it's very easy to come in these situations.

I give you very specific examples. It's very easy to say, I'm an expert in Sweden. You guys are not. I can tell you that. So I can tell you what's not going to work in Sweden. And you will trust me because I'm an expert in Sweden. Now, I'm most likely going to be wrong. Right.

Because, like, you can't really be expert in one country without knowing all the other countries and kind of see the patterns of launching country by country. Being a country-specific expert just isn't worth this much, but it's so easy to think that that's the main thing that's going to change. And so were those hires that you were making? You're like, I'm the Airbnb Japan person. Like, is that something that existed? Everybody had different, because we had multiple different jobs.

Like, the growth team's job was, like, make sure the product works, make sure the product gross, make sure people are you. Using it, we weren't in charge of talking to government. We weren't in charge in talking to press. Right. We weren't in charge of a lot of these different local things. So we always had local offices that did a lot of this work on the ground, getting host, talking to governments, talking to media. There's always people doing that that were outside of the growth team.

So we were always doing those things. But in terms of like a new startup trying to grow, grow, like those things are critical for the business. But what's critical for the growth is to really go through that kind of like step by step. that first is their product gaps? Is there some motivation in the park isn't working? Is it translated? Is there something about the growth sources that is different in these countries?

That's basically the three steps you have to go through. If all of them are true when you're not growing, well, that's a harder problem. Usually it's one of those three things. Okay. Yeah, and I would add that, especially given the three elements that Gustav lead out, for that, you don't actually need to have a growth team

in every country. You can actually do a bulk of your international growth city in San Francisco. I mean, that's what Facebook did. They launched their growth team and they had 50 million AMAUs because they were dead-vary how they were going to scale, and they needed to scale internationally. I mean, WhatsApp, the entire team is in San Francisco, but they had a majority of their user base that's outside the U.S.

So you don't really need... It's the opposite. It's the opposite. If you have a local growth team, the chance of that growth team having the same amount of experience and basically compounding effects of growth as the team in San Francisco is much smaller. Yeah. Yeah, compounding again.

So shifting to, like, thinking about growth internationally, to thinking about growth in different kinds of businesses. Steve Harrison asked, how do things differ between a B2C and a B2B company when it's going from zero to the first 1,000 users? I think it's really different depending on the type of product. I think just saying B2C versus B2B is too broad. Here's why.

You know, if you're a consumer startup, you know, it's a different type of growth approach that you use from whether you're a social network to an on-demand to a travel, right? Each of those levers tactics that you use will be different and also the time you launch will be different. If you take B2B, again, that's too broad because you have different, you know, I would put B2B in two buckets, three buckets now. So one is the consumerization of the enterprise.

You have the Slack, you have the Dropbox. They have very similar elements to a consumer startup. Now, not all growth channels are similar, even for those companies that are nuances because of the B2B nature. but you can have a growth team sooner because the adoption of those products is more consumer-driven, right, within companies, different teams and different members. The second and third, B-to-B type, actually I would categorize them in a generic bucket, which is sales, right?

Sales-driven. But there are two models. One is the Atlassian kind of model where they didn't just hire ahead of sales and plastering huge enterprise deals from day one. In fact, they figured out ways for the product to grow organically through affiliate partnerships or, you know, through their own word of mouth and website. So it was a much slower growth, if you will, at the start, but a much less spend on sales and marketing, right, as a function of revenue.

The third bucket is the enterprise sales, the typical enterprise sales, where even the first 10 users is not about getting to 1,000 users, but getting that first five large enterprise contracts, I'm talking about like $100,000, $200,000 contract. And you spending a lot of time and effort in making sure those FI really work very well and using that lens to sort of scale, right? And then you build up your sales team with the traditional sales model. So it's really different where you fall in the spectrum. I'm sure there are more nuances than the three I laid out.

But it's extremely important for you first to figure out where you fall in the spectrum. And then depending on that, the things. that you would do, whether you would invest in a growth team or not, really differ, right? So even if you take Slack and Dropbox, they did invest in growth quite early, but they also had to invest in enterprise sales. You know, you can't just grow your user base. You also have to grow your revenue. So that is a fine balance between how much you invest in growth versus how much you invest in sales. But if you take the example of a social network and a consumer, most of it is really driven by growth, right? It's all about the user and it's all about

engagement. The advertising is a form of monetization. So there, the effort has to start early. The scale of the team will be a lot bigger. This is why Facebook has a 2,000 people growth team, while Slack probably has a 30 people growth team, right? So there's a big difference in how you are. But a giant sales team. Was that? But a giant sales team at Slack. Yeah. Yeah. A giant sales. Yeah, it is. It's a big sale. I mean, it's moving more towards enterprise sales. Yeah. I think there many think they're similar. So anything from like activation to retention, how you measure those numbers. They're very similar between kind of the frameworks that Facebook came up with that growth accounting and would you apply to an

enterprise or a B2B kind of solution. You can apply a similar type of models when you're evaluating those companies. I think the big question you should ask yourself when you're working on selling to companies or enterprises is what's the level of involvement I need to have in that sale to close to sale. Like, do I need to go to their office? Do I need to meet with them a bunch of times? Well, then, growth teams isn't going to make those meetings happen faster. Now, if I just need to start in the thing about the top of the funnel, like, how many of those leads do I need?

So a new field around automating of kind of early outreach or automated sales, which is kind of have a lot of the same type of tactics that's coming from growth teams. like how do you reach a lot of relevant people with a relevant message and personalized message? Well, that's all things you can automate. And those are things that you don't really need to start a new email every single time. And there's a lot of cool things you can do there, even if you're selling to enterprises. And the same goes for like the entire kind of CRM funnel of those things. But when it comes to the actual close, the level of involvement, that's still a question that you can't just solve like through a self-serve model all the time. but I think there's going to be a lot of more things happening here

and I think we've funded a lot of companies in this space anything from sourcing to qualifying in the early part of the funnel that will help companies with sales immensely. Cool. So we did have another question and that is what is product channel fit? How would you guys define that? So this is a word I heard for the first time this week actually but maybe I should have heard of it earlier.

I think it's a brilliant definition. So what I used to say when I was at Airbnb was most large companies Google might be the exception here, grow really big through a channel. There's some platform they sit on top of when they're finding all the users. That platform could be Google. It could be Facebook. It could be Instagram. Could be app store. It could be Google Play.

But there's just some other platform that just rarely grow out of nothing. And I think it's important for companies to figure out what that platform is going to be for you. And there's a great slide by Atif from LinkedIn in one of his decks. They basically talks about how do I figure out what change. channels that I'm going to bet on. And the initial question for answering that question, what channels should I bet on is, how do people discover this product today? If you have a product with a very rare use case, you're most likely going to go to Google. Most questions that you don't ask yourself or ask your friends very often, you go to Google to find the answer,

and let's say, I want to buy flowers. You don't buy flowers that often. You usually go to Google or something like that to find that, and you end up on Yelp anyway. But that's basically how people start those kind of questions because there's not a frequent occurrence. Same with travel. The reason why Google is so powerful on travel is because we don't have an app that we go to every time to look for travel. If we travel once and twice a year, we go to Google. And we're most likely going to go to a place that we've never been before.

So we also go to Google. So that's kind of an example of how you figure out the channel that's true for you. And if you don't find that channel, you're going to rely on growth that you can't really influence that much. You're going to rely on word of mouth, which is great, but not sufficient. You're going to rely on all kinds of stuff that you just don't have as much control over. But most of these channels, you could, like SEO, AdWords, Facebook, virality, referrals. All of these different channels are things that you can actually drive yourself and make an impact. If you don't figure out which of these channels that match your product or even sales,

then you're not going to have as much opportunity to influence your growth. Yeah, and I think the interesting thing I've seen more recently, or not even that recently in the last five, seven years, is there are two or three products that have actually relied on the product to do the marketing initially. Uber and Lyft are classic examples of that. No one was actually searching online where can I find a taxi or they were looking for maybe the contact number of a taxi.

You know, you could advertise a little bit behind that. But the biggest way the word of mouth happened was actually when your friend arrived in the Uber and you were like, how did you get here? or when they were leaving a party and they said, look, I have this new service, right? So if you're doing something really cool and that's easy and convenient, then you invariably get your users to talk about it, right, which is word of mouth. You know, I've seen a similar thing which Monzo Bank does in UK. You know, it's a new kind of bank, and no one's usually looking for a bank online.

You look for it when you first want to open a bank account, maybe after school when you've graduated. What they ended up doing is they offered this really cool digital, it's a digital bank, but a digital sort of your own personal finance management tool. And so how do you advertise for that? So they send you a prepaid card, which is a glowing pink card. So imagine you're in London in the night at a bar and you take this card. Everyone's wondering what is this, right?

So, you know, in your early days, there are ways you can figure out for the product to speak for that for yourself and get that to generate word of mouth, especially if you find some of these channels are pretty hard to tap into early on. Yeah, well, if you're doing something really novel, you know, if you're like Coinbase and you get something like buy Bitcoin or you're like Vanguard and you're like buy an index fund, like there's that kind of thing. Yeah. But I could see that happen. Yeah. Cool.

So this is just going over something kind of we talked about already, but startup Sanitana asked, how do you prioritize the metrics for growth? I think this is a, it's related, but just so people know how you feel about it specifically. I would start with the funnel. Like, how does the funnel look like? like what is the first step that people do with your product was the last step. Just go through the entire funnel. Make sure you know what the drop-off rate is in every single step of them. Make sure you know which are the most important steps in that funnel. And then you kind of, after that, you'd be like, okay, which one here is like the biggest, like,

try to benchmark each of these steps. If you have a lot of channels, like do the funnel by channel and see kind of benchmark the channels as well, like what's the conversion I have for these different channels. I don't think there's a super clear answer. What we did at Airbnb is what we call opportunity assessment is basically you try to figure out everything from like what's the impact I can have on a project. What's the actual chance I think I can, like what's the likelihood that I actually hit it? How much effort is involved in kind of testing out the first version of that? it kind of depends. The outcome of that is hopefully a good prepared list in general.

Actually, that's probably where I would start. Trying to make the funnel and make the opportunity assessment. Cool. Yeah, and I think too often startups are caught in this between how many metrics is feasible versus to have too many versus too little, right? You know, I'm very analytical by nature, so I'd say you always have very few. most startups that I see actually measure less than what they should be.

But if you are working in a company that actually is not very data-driven and is going to take time for you to sort of buy into this view of using data to help and form decisions, then use the funnel framework that Gustav laid out and even prioritized on top of that. When you are young as a startup in the first and second year, really the two things that matter is the bottom of your funnel. you have lots of people coming to the website or to your app how many are actually converting to a final transaction or usage or purchase

and see the ones that are dropping off in the last two steps I would even start with the last step because that's your lowest hanging fruit because users that did not convert but went through like 80% of your funnel know about your service your product have heard so find a way to get them back right so you know and I think Alex from Facebook refers to this as the marginal user. Even when you're running growth experiments, if you're new and your growth team is new, first prioritize your churned user. They already know about you, right? So figure out why they left. Well, isn't that, that's why Shopify doesn't let you customize the checkout, right? Because

like, so many people had just like, oh, I have this new creative way for you to buy something. It's like, no, no, no, no, no, stop. Like, we figured this out over millions of purchases, right? Yeah. Yeah. Yeah. So this is another Airbnb question. question. Bradley Macommer asked, what are the two things that have driven the majority of Airbnb's growth to date? I think a lot of people would be curious. The number one driver of Airbnb's growth is Word of Mouth. So Airbnb started off with a really great product, a new concept. And Wadomoth is critical for Airbnb compared to other products because it's something that is pretty radical. Like it's not like you're going to be like the concept of having strangers come to your home or the concept of staying someone else's home for all. of us the first time we heard about it was weird. Yeah. It wasn't something that was just like natural like, yeah, I'm going to do that. So it's something we thought about for the first time, most people's initial reaction is like, I would never do that. But we all did that eventually. It just

took some time. So for that reason that is so radically different, kind of like Uber was, the first time I heard about someone getting in the car with a stranger home at night. I'm like, that's so weird. But now we all do it. Yeah. Both of those reasons, I think, those are those companies, because it's so radical and so different, word of mouth is critical because you kind of need a friend to trust you why you should start using this. Now, at some point, the brand got really big and people started hearing a baby. And before they even tried it or had a friend do it, they still trusted it. And then you can do a lot more other channels.

That was probably why word of mouth is the strongest. After that, we invested in paid marketing. We invested in referrals. We invested in SEO. All of them are sizable channels. Word of mouth is the biggest one. On the host side, word of mouth is even bigger. Oh, yeah.

That makes sense. Yeah, usually like, you know, I'm making all this money doing Airbnb. You should try it. In the early days of Airbnb, you were still paying to acquire the host, right? So it's hard to acquire hosts, like through search ads. I think on the display side, I'm not sure Airbnb did too much in the very beginning. I know Uber Lyft did a ton to acquire drivers through display ads in the beginning. So that was critical for them.

Referrals has always been strong on the guest and the host side for Airbnb, Lyft and Uber. So that's an important thing. It's kind of like just kind of a thing that, pushes you over the edge to actually tell your friend to start hosting or to start traveling at Airbnb. Performance marketing on the traveler side, though, was very critical for Airbnb in the very beginning. So people were searching for travel on Google, and they might even be searching for the travel travel, the Airbnb already was, and there was a big opportunity for Airbnb to buy keywords on Google. So that was very important in very, very early days for people kind of already knew what they were looking for.

Now, the travel, the toll keyword set of all keywords for travel is enormous. So even if we were spending a lot of time on that, like the total amount of money spent on travel keywords in the world is probably in tens of billions. And that's a very important observation, by the way, which is it's not just Airbnb. And I've, you know, seen a lot of startups in the last four years. The most successful startups have word of mother as their biggest channel for a long time. That's a sign of strong product market fit. You start paying soon.

it's, you know, it's really hard to get out of that, and it's very hard for you to continue to scale with trade marketing. Yeah. Was there a dial at Airbnb that you were using to turn up word of mouth? Is there any way to just like make a better product, give people money? We did a lot of research. Like I would say there's three moments where people talk about travel. Okay.

One is when I'm about to go to a trip. One is when I tell my friends I'm going to a trip, how are you staying, I'm saying here and be, this new thing. He came back and you came back from the trip, told me where I'm staying. that's one thing. And then when your friends are going on a trip, that's when you were like, oh, I want to tell you about this thing. I tried Airbnb. It's pretty awesome.

Those are the three moments. They don't happen every day. They're not like Uber and Lyft, but literally you're doing a trip every day and you get a chance to talk about it every day. But they do happen and it's kind of a very strong time and a powerful time where you can talk about it. I always wanted to figure out social media, but that's tricky. There are so much content posting about trips in social media. There's probably one of the top things we all post,

but to tie it back to a brand, to say that these photos you're now posting on Facebook and Instagram are tied back to the place I'm staying in. It wasn't obvious and it was pretty hard. But it's something we explored with a lot. And we've done a lot of things there, but it's something we could have done a lot more. But I think there's a lot of other companies that succeeded better with tying the experience you're having with a product in social media back to the brand. Who's that? I think Nike is a good example.

It's something that I, Nike have this thing where you're basically posting an Instagram photo with a Nike low. go that's basically watermarked on the photo. Oh, that's like when you're doing the run tracking stuff? Yeah, I'm not sure how long Nike will be able to do that. I feel like this is kind of like marketing, but that you should be paying for. Or they probably had Instagram view it. But it's something that works pretty well. And it's something that I now know that you use Nike when you went out for that run.

I don't know that I used Airbnb when I went on that trip. But I don't know which other things I used when I were experiencing that thing that I'm not posting on Instagram. That's pretty good. Any other word of mouth tips, Anu, from your research. I mean, I think that it really comes down to your product market fit. I know it sounds pretty simple. But for Airbnb, yes, it may have been conversations that you use.

But the very fact that you went to a stranger's house and stayed is something you think is cool to talk about. It's as simple as that. You're going to find a way to mention it, right? So for Uber or Lyft, it was just the convenience of which you could get arrived from your home within two minutes. So you talk about things that are 10x better. So I, you know, keep going back to, is your product really 10x better than an incumbent solution, then yes. I think e-commerce companies find this especially challenging because in the world of paid marketing, you know, everyone's trying to market the same products to the same users.

So how do you differentiate? And, you know, different companies have done things differently. So, for example, Ipsi, you know, it was founded by these three amazing, three or four amazing influencers who had a lot of following, who literally were looking. looking to them for advice on makeup and skincare. So when Ipsy launched for the first two years, they literally could use word of mouth to scale up to thousands of subscribers because of the influencers.

You know, Emily from Glaciated some similar thing. She had a blog and she did a lot of content marketing and, you know, and recommendations and advice for people on skincare before she sort of launched Glacier. And a bunch of video too. And a bunch of videos. So, right, so I think investing a lot of time up front in why you, You are that person on why you have the authority, especially if you're trying to sell a product that's quite competitive, can get you more word of mouth. I mean, Soylent.

Rob from Soilent did an amazing job of launching. Soiland, and he had lots of followers on his blog, and his first few, you know, thousands of subscribers were because of his blog, right? So I think you have to find a way to resonate with an audience outside of just performance marketing if you want to leverage word of mouth. Cool. One thing we spent time on Airbnb was basically, if you were staying in Airbnb and had three friends coming with you, capturing the identities of those people and getting the relationship with them and getting their emails, we spent a lot of time optimizing this because literally there's like three times many people that were already booking with Airbnb that could be our audience. And we just made some math on like where would that be in 2020 and that would be like half of the world. It was basically looking at everyone's adventure. Eventually you're going to spend some time in an Airbnb.

If you think about your parents and everyone else that you're booking for that you otherwise might not have. use the service to make sure that they know that this is an Airbnb and then they have an account. I think those are some of the things we spent time on. Anything else you want to cover? No, I think we pretty much touched. Thank you. Yeah, thank you. Yeah, this is great.

Thanks, guys. Yeah, thank you very much. All right, thanks for listening. So, as always, the video and transcript are at blog.combinator.com. And if you have a second, please subscribe and review the show. All right. See you next week.