Lessons Learned Growing Successful Marketplaces w/ Mike Duboe, Lenny Rachitsky, & Dan Hockenmaier

Hi, everyone. I'm Mike DeBeau at Greylock Partners. Welcome to our podcast Gray Matter, where we bring in some of today's top entrepreneurs and business leaders to share their stories from startup to scale up. Today we're talking with Lenny Rachycki, former growth leader at Airbnb and Dan Hockenmeyer, founder at Basis One and former head of growth at Thumbtack. Great to have you here, guys.

Great to be here. Thanks, Mike.

So this is the third of our discussions on growth this year. We've previously done a deep dive on product and growth fundamentals with Brian Balfour and Sean Clouse. Then one more, marketing and measurement with Tyler Elliston and Kim Larson. I'm excited for today's conversation, given both guests have come from Marketplace backgrounds and now have a breadth of exposure as advisors and consultants around growth. Lenny and Dan, I know on a personal level, are the two of the sharpest, kindest, most thoughtful folks working on growth. And so before jumping in, we'd love to just get the audience acclimated and spend a little bit of time on your background. So, Dan, tell us a bit about Basis One and your previous role at Thumbtack and maybe why you decided to jump over to the consulting side.

Absolutely. So I founded Basis One completely on accident. I had been coming off four incredible years at Thumbtack. I joined what was about 30 people, got to help build the team to 500 and build a bunch of pieces of growth and other teams along the way. So I started working with founders and CEOs, mostly as a way to evaluate what was next and was getting this incredible consistency of types of questions they were asking around how to prioritize growth opportunities, how to unpack customer retention and monetization. And because these were kind of messy problems that were both quantitative and qualitative, I started pulling in people I'd worked with before, so analytics folks, user researchers, growth people. And so we looked back and realized we were building a consulting firm on accident and it was working really well. And so we decided to formalize it and have been fortunate to work with a bunch of great growth stage consumer and SMB companies. So looking forward to doing more of that.

Lenny, you know, you just came off a pretty epic run at Airbnb, spending seven years there with a host of roles across growth or product in some capacity. You also came from an engineering background, so tell us a little bit about your journey.

Yeah, I was a computer guy from maybe age 12. My parents got me a computer. I think it was like a 386 with like the turbo button. Remember that? No, maybe not. And so I went to school for computer science in San Diego, got a job right out of college, joined a startup, stayed there for a number of years, LED engineering, did some R and D. And then at some point I decided I wanted to start my own company. And I moved to Montreal to do that, to make it even more challenging. I had some friends there that were starting an incubator program and just kind of convinced me to start it there and funded me and helped me out. And so moved up there, started this company. It launched at south by. We did pretty well, got some funding, moved to San Francisco, got acquired by Airbnb about a year later and spent seven years there. And I left Airbnb about eight months ago at this point.

And what were some of the. I know we could talk more about your experiences at Airbnb, but the roles that you held over time there.

Yeah, so I came in as an engineer. As an engineer my entire life. And then I moved into product about six months in, I was the CEO of our startup. And so when I joined, I was kind of doing PME things, and our head of product was like, what are you doing? You're a product manager. And so I moved into that. And so we started working initially on a. On a way for guests to get together in a city and. And to do things together to kind of make travel on Airbnb a little more social. And so we built that and launched in San Francisco, and it was doing okay, but then there was kind of shift in strategy, and as often happens with acquisitions, we decided to kibosh it and move on to something else. So that never launched. And then that morphed into working on helping the host community become an actual community by building a way for them to communicate and share stories and tips and tricks. So we built that, that evolved into a kind of what we call the host team, which. Where we basically built tools for our hosts to be successful and have everything they need to be hosts. That morphed into eventually into Instant Book, which was a way for hosts to accept guests instantly. And I could talk about that for three hours. Worked on that for two and a half years. That was a big investment. Airbnb and then worked on booking conversion, helped launch the superhost program, helped build out the supply growth team. Worked on that for two and a half years. Also reviews a lot of different stuff.

Yes, this is a good transition into the first topic I wanted to spend time on, which is just the definition of growth and org design around It I think, you know, there's a lot of companies that are looking to spin up a growth team for the first time or are looking to evolve some of their current product or marketing teams to growth teams. I look at Airbnb as an example, knowing some of the folks that worked on growth there, as an example of a company that seemed to have got a lot right. And so tell us about what growth. Org looked like at Airbnb and maybe related to that how you define growth personally.

Sure. The way I think about growth is most companies, especially if they're venture backed, they need to grow to be successful. And so in a sense, everybody should be working on growth one way or the other, whether they're called growth team or not. The way we thought about growth in the end just kind of very concretely is just is revenue increasing? The way we measured it is bookings, bookings per year. And so yeah, so at Airbnb we historically never had a growth team. It was just do what we need to do to help host and guest travel and be successful at some point. Gustav Ahlstroimer, who's now partneryc, basically kind of built the initial growth team, kind of realized, hey, there's best practices out there for how to grow a company. Let's figure out how we're growing today, let's figure out how to move levers that are working, let's figure out how to measure. So a big transformation at BNP was even understanding what was incremental growth versus just like random luck and word of mouth and strong product market fit. And fast forwarding to today, there's very clear growth teams on supply and demand that are basically their entire job is grow.

And how does the cascading of metrics like you mentioned, revenue is kind of the top line metric, then that kind of cascades down into some host metrics and guest metrics specifically. But if you look at today, what is the accountability and hierarchy of metrics look like on growth?

There's so it was actually not even revenue. We actually just looked at bookings. That was kind of the final metric that we all thought about. And then we thought about what are the levers that grow bookings, more supply, more demand, things like that. And so we formed teams around eventually we started calling them input and output metrics. So in a sense we had these input metrics like supply, like conversion, like traffic, and build teams around that. And their job was drive that metric however you want.

One of the things I always found fascinating about Airbnb is in talking with Gustav early on was many companies will separate growth functions into either product led or marketing led growth, and the composition and DNA of the team will look like a result of what that top line decision is. I think at Airbnb, my outsider's perspective is that you actually applied a lot of product DNA around doing paid marketing. And so those teams actually looked quite technical. Talk about that a little bit.

Yeah. There was a very stark change when we moved performance marketing into the product Org out of the marketing Org. And I think there's a lot of reasons for why that was. And I don't know if this is the same at every company and how if it works better if it's within marketing at other places. But what we found is so much of growth and marketing is productized now, led with data and engineers building infrastructure to automate it and not kind of wing it. So the way we ended up moving things was the performance marketing team is just another team within say the guest growth team or the host growth team with a PM leading the team, with marketers on the team, data scientists, engineers and things like that. And that's worked really well for us.

Yeah, I think Thumbtack took a very similar pattern. And one thing I'll add for companies thinking about this is customer acquisition, particularly until you get to some level of scale, call it 100 million revenue, is mostly about getting really good at one thing. And so it's less about where the team sits and more about giving that team all of the levers to make that work, whether it's virality or paid marketing or SEO, which it was in the early days on the consumer side at Thumbtack. And none of these are any longer about just marketing, I think like even paid marketing, to your example, if you look at like, who I think are some of the very best at this right now, if you look at like Wish, for example, that's an engineering team, not a marketing team. And if it's a purely marketing effort, you're bringing a knife to a gunfight effectively now, I think. And so it's a. It's a very different game for sure.

Yeah, it's something we used to talk about a lot at Stitch Fix. I think right now for the usual suspect channels, your ability to go and find arbitrage or actually differentiate on being great at paid marketing. It's actually really difficult and if not impossible, just given how efficient these ad marketplaces are. And so we would say that running great performance marketing is a function of having great creative and generating that with enough frequency. And then also Having great data pipelines, which is actually more of a product technical data science problem. So we try to leverage data science very heavily around what we did there, specifically around measurement and get smarter around that. But I definitely think that more of the DNA that runs these performance marketing functions will start to look more technical and more like product teams. And it's something Dan, I mean, you and I have talked about, we talk about at Reforge is, you know, the field of growth, if successful, might cease to exist or might abstract itself away into product and marketing functions if it's done right. Like, do you see that happening already today?

Yeah, certainly. So I think one thing that's really important for companies to understand is look at where the function of growth was incubated. It was Facebook and LinkedIn and Twitter and then Pinterest and Airbnb and all of these companies. And there's something that's true, that's common about all of those, which is they have extreme product market fit and they have network effects or compounding benefits from small changes. So in that world, you can have a pretty traditional team that is data led, experimentation led. You can avoid a local maximum problem because you're pretty sure you're tapping into the right vein and you can just continue to test. And so if you're working on that kind of business, Airbnb, thumbtack, some of the businesses we're working with now, like fair, very much follow this pattern. And for people who are in the growth world, those are really fun places to work. But for other businesses, actually not always the best fit. And in those cases, my hope is that they take some of the best things about growth. So still being data driven, experimentation focused, but combine it into a more traditional product or marketing process. And so I think part of what we're seeing is companies realizing that they can't actually just apply this playbook to a different type of business.

Something I'll add to that is the way we thought about building teams at Airbnb, especially thinking about growth is so say we had the host team, we kind of divided things up into host and guest is growth was a part of the team. But there was also just like, we want to make the product experience better for hosts, we want to make hosts happier. That's important to us. And so there was kind of a growth component and there was a. We just believe in these bets, like the superhost program, for example. And what I found really important is that it's all under one leader that can trade off against one versus the other. Like we need to let's put all these resources behind something we believe in versus let's focus big on growth versus there's kind of competing teams siloed and separated because the trade offs are much more complicated and they're kind of battling. It's nice for somebody in either a leader or a leadership group to make those trade offs amongst themselves.

Is the implication of that in your mind, that there should not be a growth leader reporting to the CEO, but growth should roll up to something like a product leader who can break those ties.

That's the way we approached it. So the leader of the host team wasn't like, they're just growth. They're responsible for the host on Airbnb, and so they're responsible for getting more hosts. They're also responsible for their general happiness and satisfaction and all those things. So they were able to have this wider perspective on what's important now and where do we want to put resources.

Yeah, I mean, for what it's worth, Thumbtack followed a very similar pattern.

I think it worked well where we.

Had some functions originally under marketing, some under product, and ultimately it all came under product and it was really effective in that way.

We're talking here about companies that this is an evolution that was made at scale at Airbnb. I think when you look at companies, maybe at the Series A or closer to inception, I still think I am seeing more frequently than not there being standalone growth pods that are generally designed to move faster than the average pace of the company and are measured more in velocity of experimentation. And then the stuff that works will get rolled into core product. But that dynamic actually breaks at scale. So I'm curious, if you, like, when you work with earlier stage companies, do you see that? What trends are you seeing around that point?

So I think one helpful frame is success is measured by velocity of learning, not just velocity of shipping experiments. The implication of that is that you actually in some cases may learn from shipping something bigger without an experiment, or in some cases launching fewer tests that are more holistic tests. And so it's funny, we see two types of problems. You see early stage companies who are not experimenting at all, and they should be, or they get to a point where they're later stage and they're just testing everything and all they care about is velocity and they're using it as a place for critical thinking. They're replacing actually thinking hard about who the customer is. I think you want to be somewhere in the middle of those two.

This is actually a topic I wanted to get at later. But while we're on it now, around analytics, something that was fascinating to me when I spent some time at Airbnb was looking at how democratized the experiment framework. And actually the analytics infrastructure was basically anyone, it seemed, could go and kind of spin up an idea and actually look in real time what the results were of that. And that, I think, has benefits culturally, I would imagine, to an extent. And one of the paths that I have seen, maybe as kind of a watch out for companies that start to do growth early, is testing changes that are too micro, that even if they accumulate, aren't really going to move the needle much. And so I'd be curious how the question's more on the experiment framework and the democratization of that kind of how that panned out at Airbnb.

Yeah, so the way that it started was a product or a library, I guess, called Trebuchet, that was essentially a flag that basically assigns half of users to one group and half of users to another. And in code, you wrap your change around this if statement. If they're in this trebuchet, show this. Otherwise show that. And that was all you needed to do in code. And then it sets a cookie and remembers you're part of that experimental group. And then it was just like a bunch of SQL queries to figure out what happened, what happened between these two groups. So it was very manual, but it gave us some way to start tracking this stuff. And then they ended up building dashboards that let you, like, turn them on, turn them off, launch to 100%. A problem we ran into is the code base just started having all these remnant if statements of experiments that launched, but no one got back to cleaning up the code. So that's where it started. It was really simple and it worked really well. But then we realized there's experimental issues with that. You're diluting your experimental group pretty significantly because you're not assigning it at the intervention. You're just kind of assuming everybody saw what you're experimenting with. So there came a point at which we put a team on building our own internal experimental framework, which we called erf, Experimental Reporting Framework, which was in code similar to that, where you kind of wrap changes around this if statement, but it smartly assigned you to an experiment. And then to your point, it created these dashboards where you could watch your experiment all day. I had like 30 tabs always open in a. In a browser window that I refreshed every day that I went through to see how our experiments were going. And it was, you know, it shows you if it's statistically significant, what the P values are, how many people have been assigned to it. Importantly to had core metrics we call them, which were important metrics to the business. So that even if your experiment isn't impacting that or even you don't care about say review quality, you'll see that in your experiment and make sure you're not hurting that. And it lets you look at all experiments pivoting on a metric so you can see what experiments are most helping, review quality or most hurting bookings. And then it could email you reports of how your experiments are doing, let you see experiments by team.

One thing I'll add is we see a lot of the companies we're working with, particularly ones that were founded in the last call it three to four years, even if they have incredible engineering and data science teams are using third party tools for a B testing. Increasingly the full stack tools are quite good because we had a similar homegrown system at thumbtack. And then you run into issues where if you're not constantly running AA test you can miss like bugs in the code and other things. And so even very smart teams until they get to a certain scale are relying on third party tools.

Yeah, it's like a massive investment and the team working on it is dedicated to this. And it's. I don't even know how many engineers, data scientists and they're always uncovering flaws that they didn't know about.

Yeah.

So we've talked about the deployment and measurement and ongoing kind of launching of experiments. How about on the like as the head of growth or whoever's in charge of the function prioritization and determining actually what to launch. Anything you guys have seen as best practice on just prioritization of experiments?

Yeah, I mean I think there's a couple levels to this. One is what's the metric or loop that you're working on so fundamentally like what's the thing that you care about and making sure that you're testing into that lane. And this is at the level of should we be thinking about activation and retention versus acquisition. And I think they're very important distinctions to make there underneath that. I think one of the things I'm a big fan of going back to thinking about what optimizes learning is organizing all the things you could ship by hypothesis. It's testing so that you aren't making a bunch of runs at answering the same question, but you rather start to learn what's true. So if in a Conversion flow, you might be testing things around trust, things around enough information, things around tactical things like payment flows. And so making sure you're hitting each one of those and answering the question before you realize whether or not you should move on is really important to the kind of overall prioritization function.

This is getting to a really important point around growth models. And so I want to change gears and talk about that as a concept because I think a lot of companies will start to ideate and get excited about ideas to test and actually start to move the needle on yet I think what's more important at a meta level is figuring out what is the right needle to actually move and how to prioritize on working on one stage of the funnel versus another or one potential user group versus another. And I think the most fundamental piece to have in place before you do that is a growth model yet I think as a concept, high level, it's pretty widely understood what output from something like that could look like. But as far as actually the mechanics of going and building one of those, I still think it's fairly uncommon. And so, Dan, I know it's something you with your practice have spent a lot of time on. So I would love to hear, I guess when you're stepping into a company before they actually have the fundamental growth model put together, yet might already be doing things like where do you start? And walk us through kind of what an ideal growth model setup could look like.

I'm excited to hear this too.

Yeah. What a growth model needs to accomplish is an understanding of how the business grows, how customers flow through the product, how they generate revenue, and importantly how that drives a self sustaining loop. And I know loop is this very buzzwordy kind of topic in growth right now, but all it means is what's the asset that it's generating as a byproduct and how do you reinvest that asset into more growth? If it's for reality, it's pretty obvious users create more users and so on. And you can measure the K factor. If it's paid marketing, it's monetizing users and reinvesting that. And this is why it's really important to think about something like payback period instead of LTV to cac, because payback period literally measures how fast that loop can turn. Content is another one where users are generating content which you can use as a net. And so to your question on how you start this, there's the qualitative version which is literally just asking people, do they understand how the business grows. And I am constantly amazed. If you ask five people at a company, you will often get five answers. And so this qualitative step is super important. What's the central loop or set of loops that's driving the business? I do think it's really helpful to then get it into an actual analytical model. And it's amazing how big businesses get before they move off of something like Google Sheets. So it's okay if this is like pretty basic spreadsheet model which describes acquisition channels, how they convert and activate, how they retain, and then how those cohorts generate these byproducts. And I think from there you can start to see all these really interesting things which aren't so super obvious. So for example, at thumbtack we went through this exercise too late and I think as a result paid a lot more attention to activation and conversion for too long. And you know, it's kind of obvious that you should focus on customer retention, at least theoretically. But it's hard to understand magnitude because things like as you increase retention, users become more valuable, you can go reinvest more in paid marketing, they generate more content, which increases SEO. So the magnitude is much bigger than you would guess on a kind of linear basis. You also start to see the actual mechanism. So it turns out all of this repeat engagement was happening on mobile. So it's really important to push people from mobile web over to native. You see certain segments emerging. So once you get it into a model in this way, you can start to really piece apart what's happening. When we build this for companies now, it's a weekly model that's usually on a daily basis pulling in data automatically from whatever their data warehouse is. And it's doing two things. It's one, it's projecting out in the future so you can use it for opportunity prioritization. But it also can power like a weekly growth metrics review where you say, here's the 80 metrics we care about. There's like three or four top level ones on any given week. These five of them are outside two standard deviations where we would expect them to be. And so we should pay attention and figure out what's happening with those. So it has both this kind of forward looking and backward looking component.

So let's shift gears and talk about marketplaces a little bit. I think it's an area, Lenny, I know you were working on a broad study of marketplace models that by the time this podcast is released will have started to drip out. And so I was excited to see some of the work you put together on that, you know, what were some of the most insightful lessons you learned in studying both early and at scale businesses?

Yeah. So I guess a little bit of background, I've embarked on this, what turned out to be a very complicated and large investment of research into how the biggest marketplaces today grew, how they cracked the chicken and egg problem, how they grew, supply grew, demand maintained quality, things like that. So I ended up interviewing about 24 people across 17 different companies to figure out what worked for them, because I felt like just the one perspective from Airbnb was not that useful. And so that turned into this, I don't know, 50 page long research document that I'm going to be releasing little bit by little bit because it's way too much information. But I learned a lot from it and I'm hoping it's useful to other people. So a couple things that I was a little surprised by and we'll see, we'll see if other people are too. One is. So there's kind of four steps to cracking the chicken and egg problem based on the talks and interviews I've been doing. One is you figure out how to constrain your market, whether it's geographically or category. Two, you figure out if you want to focus on supply or demand, which one you want to concentrate on. Three, you grow supply. Four, you grow demand. And so diving into kind of how these worked, I'd say one surprise is that almost 80% of marketplaces focused on supply. And there's only three that I talked to you that concentrate on demand. And they all had very kind of unique reasons to do that. So learning there is, unless supply is coming extremely easily, almost surely you should just be focused on supply.

What are the examples where supply was coming easily? Because I think that, I think those of us that talk more about it, that might be intuitive, but I think to many out there it's actually counterintuitive. And so give us some examples of the few that focused on demand first and why.

Of the 17 companies I talked to, three that where supply came super easily and essentially they were never supply constrained. That's kind of the message I got from those guys were Rover, TaskRabbit and Zillow. They're all unique, but Zillow is unique in that they basically bootstrapped supply by using existing data that existed around homes and then built, I believe it was the zestimate on top of that. And so they didn't have to convince people to get online initially. They created this kind of bootstrap supply. And then they created demand. And then they went out to people and convinced them, hey, you should list your home on Zillow. Because there's all this demand for Rover and TaskRabbit. I think the key for them is the value prop is really great. Make meaningful money anytime you want by walking dogs or putting together furniture. So anybody could do it super flexible. You make meaningful money. And I think that's a rare thing. And so Those are the three where supply was not a problem. And TaskRabbit's a good example where they actually had to end up charging a fee to even apply to be a tasker because it's so much backlog.

Yeah.

So, Dan, how about on your end when you were scaling the marketplace at thumbtack to different levels, what were some maybe non intuitive insights you kind of came across over time as far as marketplace learnings?

In the example you gave Lenny, you talked about how most marketplaces focused. Thumbtack was the example of the extreme opposite, where the founders decided to launch in a thousand categories nationwide.

For good reason, though. That's right.

And they were also just, they did some incredibly scrappy things that early on they just had a lot of hustle that allowed them to get past that hurdle. But I'd say it went in phases. And in the early days, the growth team would kind of like pivot from working entirely on supply to working entirely on demand, depending on what was needed. And for example, like when I joined, we had a burning supply problem. SEO was starting to work on demand, and we had massive influx of consumers with actually a very bad NPS because we didn't have pros for them. And so it was all hands on deck. How do we scale that? But I will say I think the key insight that guided all of this work was understanding how we define liquidity and goaling both sides of growth on that metric. And I think liquidity is one of these things where prior to getting there, it's basically all a marketplace should focus on, because the marketplace who hits liquidity first typically wins in that local market. And that typically means winner take all or winner take most. My personal definition of liquidity is when the marketplace becomes reliable for both sides of the market, when they can rely on it to either generate business or hire somebody. There's one exercise we went through of how does the market break down? In thumbtack's case, it's pretty obvious that it's by city and Geo and the units of supply and demand are not really permeable on those lines. That's the unit at which you need to build liquidity. It's made a little messier if you look at a commerce marketplace like Amazon, where one person might want to buy many things, but you need to get kind of like down to the canonical unit of the market, and then within that unit, what's the metric? That means you're doing a good job. In my mind, the forcing function is always is demand happy? So are you doing the kinds of things that will allow them to convert at a high rate? In our case, or in Thumbtack's case, if you got them enough quotes from pros, there was a point where they tipped, they started hiring, their NPS spiked, they would retain at a really high rate. We literally built this huge heat map that had the rate or the percentage of consumers that were getting enough quotes in each market. And we said at above a certain threshold, it was healthy or not. We actually plugged this into all of our growth and marketing programmatically. Bidding was going up and down in markets for supply and demand based on where that liquidity metric was. That's where it really started to click. And we stopped being reactive and jumping from one side to the other and really were thinking about how to manage liquidity going forward.

Yeah, it's insightful. Lenny, how about liquidity? What metrics were using to assess liquidity at Airbnb?

Yeah, and this is one of the things I looked into with this research is how to one, how to marketplaces know if they're supply or demand constrained. And a couple learnings came out of that. One is something like 40% of marketplaces that I spoke with were always supply constrained. 40% were one or the other and depends on market or category. And then 20% were always demand constrained, which is, again, really rare. And so of the companies that were always that were kind of in the middle, that wasn't really clear, they essentially always came up with a model that kind of told them where they're supply constrained. At Airbnb, honestly, I think they're still trying to figure it out because it's very complicated. But kind of the evolution that we went through is initially it was. We kind of defined an occupancy rate in a market. If it was above a certain percentage, we thought that we consider that supply constrained. And so we went from like maybe 40% to 70%. It's definitely not 100, because all the good stuff's taken at some point, and what's left, people don't really want to book. And then we pivoted A little bit to thinking about. If you look at a chart of conversion overlaid with occupancy, you should see a point at which conversion drops if occupancy exceeds some threshold. And so that started telling us that in that market that's supply constrained. And so that became a way we measured where we need supply. And now I think they've moved to kind of an econometric model that tells you for a unit of supply or demand in a market, how much revenue do we think it's going to drive? And I think that's how they think about this stuff now. It's much more complicated.

One of the other areas around marketplaces that I've personally been spending time that I think is becoming a more interesting category is around B2B marketplaces versus B2C. And I know a lot of the mechanics are very similar between the two. I'm curious in your studies, Lenny, if you spent time with any B2B marketplaces, what you saw as being applicable or not in that domain that looked different than on the consumer side?

What's interesting is in this research I kind of discovered there aren't any massive B2B marketplaces other than Alibaba. That's the one that came to mind. So either I just haven't heard of them or there's just getting started. Like Fair comes to mind as a good example.

Fair's one. I think we're starting to see some in the trucking space. I think it's certainly a later wave that's coming now. And there's some really interesting ones. Having spent a little bit of time on these, I totally agree. Actually, the dynamics are strikingly similar in a way that I think people probably thought would not be true. But some differences are. One, the cross side referrals can be even more powerful because you can, instead of just asking somebody to go out of their way to refer to, you can just work it into the natural course of business. If you look at ritual, for example, most of their acquisition comes from allowing bars and restaurants to use the product for free with their existing customers. In the course of just doing business with them, they're exposing all of their customers to the demand side of the platform. There's many opportunities to do that kind of thing. One other thing that comes up that's pretty interesting is there's something different about the word of mouth dynamic in a B2B market where people are more motivated to seek out services to grow their business and there's more established channels to talk about a business. And so like we saw this at Thumbtack, where in the early days we were spending a ton of money on supply acquisition. But at some point awareness in the market tipped and we just started getting an avalanche of organic demand because people just knew about it and it was part of what they researched. And I think B2B businesses will see this more often, which, you know, the plus side of that is you can probably be more efficient with acquisition. The negative side is you now have a community to manage and actually like your message, messaging is a little harder to control because it's happening in established forums where they're talking about you. I think there's a couple of wrinkles for how B2B businesses go to market.

I also think you see the evolution of many of these B2B marketplaces becoming fintech businesses earlier in their scale maybe than on the B2C side, which I think is interesting as well.

That's fair.

What's interesting about that is another thing that's emerged from this research is if you look at which levers are most common for demand across marketplaces early on, number one is word of mouth. Two is the supply driving the demand. So like doordash signs in a store, people downloading the app because they want that food, or Etsy is an interesting example where the supply buys things also on Etsy. And that became kind of. They're almost the same person. And so it's interesting to hear in B2B. That's also true. Sounds like you said both word of mouth and the supply driving the demand.

So this is a good point. It's not just cross side network effects, but it's actually supply becoming demand and vice versa. But if you look at like change.org, a petition that goes out to millions of people or eventbrite an event that millions of people see, or maybe fewer than that in that case. But you have this massive exposure on the demand side and when you start to model this and stitch together email addresses or some other identifier, you see that a bunch of them are becoming supply at a much higher rate than you would expect, naturally. And then there's all these things you can do to influence that around how you message it, email and retargeting and other things at certain points. So this is a super powerful lever for marketing marketplaces.

Yeah, the last piece, and I want to move on from marketplaces in a moment, but I think disintermediation is of course a popular topic that comes up. And while we have Dan here, I think thumbtacks, you know, an example, I would imagine that has the potential to struggle with this more than others, or at least in the early days, like how did you guys eventually think about defensibility against disintermediation and how does that apply to businesses you're looking at now? That should start thinking about that earlier.

I think it's a great question. My general philosophy on this is if you have to do too much to police your marketplace, you don't have great product market fit. You've got to create something which makes it so much better to transact and that value you're delivering is so much higher than the price of doing business on the platform that they want to transact anyway. In Thumbtack's case, we knew that was happening. We did price it explicitly. So there were cases where we would charge more for a recurring home cleaning lead than for somebody who's looking for a one time cleaner because there was going to be an established relationship that came out of that that we knew might go off the platform. But in most marketplaces, I think it's more about thinking about growth and monetization and product strategy to actually make disintermediation not a problem.

Yeah, at Airbnb, I think it's rare that you stay at the same place again. And so, you know, you find it in Airbnb. You're booked through Airbnb because I guess a few things we did. One is we had this gnarly regex that looked for anything that looked like a phone number and email address and increasingly became more and more complicated to make it very difficult to even exchange contact information on the platform. And two, I think the guarantee, it was super effective of the insurance policy on the host side where you get a free insurance, million dollar insurance policy for using it. And that's worth like 3%. It's worth paying 3%, which is basically the credit card transaction fee to the host.

Yeah, I'm seeing that more commonly as well, like insurance and other financial products as being a key protector there. All right, guys, so the last point on marketplaces, I think another kind of trend that I have seen more that our friend Casey Winters just wrote about was this business model we're seeing of businesses that start as SaaS, companies evolving their model to a marketplace model over time. Kasey has some great thoughts on why this is actually pretty atypical for that to work. Lenny, I know you and I were just chatting about this earlier. What thoughts do you have on that topic?

I guess, first of all, if anyone's interested in this they should just go read Casey's post. Just Google Casey Winters and you'll find his site. And I think it's the most recent post. But something that comes up in this is through the research that I was doing, the story of Patreon, where I thought they were a marketplace. I thought they drove demand to creators. Turns out the founders thought it was a marketplace and they kind of operated as a marketplace. And what I've learned is they, they kind of had this realization that maybe we're not a marketplace. What if we just cut off the entire demand side of our roadmap and just focus on becoming the best SaaS tool for creators? And what they found, one, the way they, they put it is it was maybe the, one of the best decisions they ever made. And two, it aligned incentives between the creator and their company because what they found is that once a creator gets too big, they almost don't want to be on the platform anymore because they've become this kind of channel to drive growth for other folks versus for themselves. And they almost don't want that. And so they call it the graduation problem. I think a lot of times people want to become a marketplace. I think one of the more interesting realizations sometimes you're better off not. And so I think a lot of companies are going through that.

I think folks who have worked on marketplace businesses realize how hard it is to aggregate demand and why there aren't marketplaces built in a lot of these spaces. And so it's easy to say we're just going to go kind of do that thing. But if you haven't done that in a big way so far, it's often hard to make that transition. I think the exception is when SaaS is used explicitly to bootstrap a marketplace. So it's like the goal in mind versus something where you have an established SaaS business that you then try to layer a marketplace onto. I think it's a harder pattern.

It goes back to some of what we discussed earlier about building the fundamental growth models and the metrics that you care about as a marketplace business versus SaaS company. It's a pretty big, almost refactoring. When you do make that evolution, and if you're deliberate about that, you could see it working. But culturally, that's another factor that's worth noting as well.

Yeah, the way I think about this stuff is your supply may want more demand and you may want to offer them more demand, but if the demand doesn't want what you're offering, it's not going to work. And so the most important part is just do people want this? Is there demand out there and are you solving a problem for people? Because if you're not, then you're never going to turn into that marketplace. And to the point you made about how you almost have to think about it ahead of time and that's when it works out. Well, I think OpenTable is a really great example of that where they kind of knew they were moving into this two sided marketplace, but they knew they didn't have any demand yet. And so their whole what I learned from some of these interviews is their pitch initially was 90% this is going to be the best reservation system, replace your manual books that you kind of lose and pass around. And so just use that and that's it. And 10% of the pitch was in the future we will drive you demand. And as they built up the supply and started driving demand, they shifted their sales pitch little by little and now it's mostly we're going to drive you a bunch of demand.

All right, so that was a good preview for Lenny's Marketplace manifesto. If you want to read more about that, subscribe to Lenny's newsletter. So we have time for a couple more points to hit. And one of the ones I want to hit is just misconceptions about growth. So common mistakes or failure modes. I think it's something that as this topic of growth becomes more popular and more companies want to spin up growth teams, I think there's probably more examples of getting it wrong than getting it right. And so I think for both of you guys, like what common mistakes and pitfalls have you seen companies make as they're starting to think about spinning up this functions and maybe as an extension of that, like how do you see the evolution of this function in the near term?

I'll take that first. A few come to mind. I think one classic one is just this idea that there's going to be some growth team over here and they're going to take care of growth and we can work on like stuff we really want to work on, like make amazing products and make everybody happy. And I think in reality it doesn't ever work that way. I think those have to be super connected. And so don't expect some kind of magical growth team that just solves all the hard problems for you and you can work on beautiful things off to the side. Another is that it needs to be short term. I think a lot of people think of growth as like this incremental optimization. That's all it is. And the way like all growth is, is you pick. Here's the thing, we want to grow. Let's put smart people behind that problem and let them figure out how to grow. And it could be conversion, it could be quality, it could be happiness of your community. And so that's all growth. The key is, what are you growing and what resources are you giving the team to solve that problem?

All right, so last topic to talk about is for companies that are starting to think about measuring their growth and who are starting to invest in it, especially in this world of proliferation of capital and there being many ways to actually go and spend on growth, some of it potentially being unhealthy, how should a company think about whether they are growing in a healthy way or not? And maybe for ones that aren't, what options do they have?

Yeah, so I think about a hierarchy of the quality of sources of growth. And so probably the most healthy growth is purely organic word of mouth. A lot of this is going to show up as traffic or customers that you can't track because you don't know where they came from. There would be, then, I think, channels you're using that are proprietary to you in some way. So it's using your own content to drive SEO, it's incentivizing your own customers to drive referrals, those types of things. And then there's totally non proprietary channels like paid marketing. If you think about what the business looks like, this goes back to the growth model. You can literally just strip away the third tier or the third and the second tier and see what you have left. Is the business still growing? Is there still something underlying that's valuable? If not, that's a difficult situation to be in because I think things like paid marketing are typically best as accelerants rather than as the core growth strategy in a marketplace. Great ways to launch a market, great ways to get a loop spinning. But if your business is relying on it to the point where you take that out and the business doesn't look like it has a lot of growth in it, it's probably unhealthy the way.

I guess I'll take a different approach to this question of how do you build a business and a community and a company that works out and makes all your kind of stakeholders happy and doesn't just focus on growth at all costs? So at Airbnb, I think there's this always this spectrum of focus, like what's a metric that we're driving, like, let's say bookings. And there's a lot of power to just a lot of focus to, okay, but there's all these other things that matter. Are people having a good time, our communities happy, our hosts satisfied? Are people creating good memories and things like that? And so Airbnb kind of went from one end of the spectrum of we're going to grow nights booked, and that's going to be the focus and we'll see how it goes. And I think, honestly, I think Brian and the founders were early to realizing there's much more that should be put into that equation. And so there's this focus on what Brian ended up calling a 21st century company. As we go into the future, we're going to realize one that there's a lot more stakeholders we have to think about. And so with Airbnb, it's the shareholders, but it's also the host and the guest and the communities. And so that's kind of become more institutionalized at Airbnb and they're actually starting to measure some of these things. And then two, there's this idea of an infinite horizon where you should be thinking about your company as it'll be around for 100 years, and how do you, how do you get to that and how do you plan for that versus the next quarter, the next year? Airbnb has gone through that evolution from just this one metric to this broad bucket of metrics.

Yeah. The perspective I would add, actually, from my time at Stitch Fix is I think some companies, and I see this as an investor, sometimes companies that really bootstrap growth early on with paid marketing and then are. Bootstrapping is probably the wrong term there. But get growth moving with paid marketing early on and then saying we're going to introduce organic late in the game. I've rarely if I've ever seen that work because it's a very different discipline. And it's just, you know, that is dangerous when it doesn't work out. And unless you have the economics to make your CACs work at much worse levels than they look, when you're early stage, those games typically don't end well. And so I think Stitch Fix was a good example of a company that the early days, because of constraints actually around capital, we were focused on a particular demographic that really had strong word of mouth and just amplified what this concept was to their communities and networks early on. And so we actually didn't need to do anything or couldn't do much on paid marketing also because of supply constraints. And so by the time we started layering that on, we had a really strong foundation of retention, a really strong word of mouth component that actually had a K factor on every user we acquired. And so when we started to do it, it was very much an accelerant versus a crutch that was a recipe for healthier growth. Even when you get into tactics like paid marketing that can on the surface look unhealthy. So, yeah, the takeaway is there's no one size fits all answer around this stuff or any of this stuff. But there are certain patterns that could start to smell very suspicious at the start.

One addition I'll make to that is if you're in a business that has real network effects, it's the only case where your ROI on something like paid marketing may actually get better over time. That's why you see these businesses sometimes spending what looks like too much in the early days, but for almost everyone else, it's likely to get harder. If it doesn't work now, you're in a tough spot. Thinking about where you are on that degradation curve is really important.

So, you guys, this has been great. We could definitely have a continuation of this another time, but I want to wrap things up here and so thank you both for your insight. For folks that have listening that want to hear more from you, where could they find each of you guys?

So I'm on Twitter Dan Hockenmeyer, where I tweet not very frequently and basis one is at basis one spelled out dot com.

I'm Lenny San with an S a N on Twitter and I have a newsletter that I've been working on@just lennyrichtsky.com or lenny.substack.com awesome.

Thanks, guys.

Thanks, Mike.