Andrew Chen on finding the “fresh powder” in growth

What do Dropbox, Uber, AngelList, Front, Gusto and Boba Guys have in common? All have benefited from the sage advice of growth expert and Andreessen Horowitz general partner Andrew Chen.

Andrew’s been an angel investor and advisor for a slew of name-brand startups; however, he’s most widely known for his invaluable essays on growth. He’s written more than 650 of them over the past decade and has been featured and quoted in The New York Times, Fortune, Wired and Wall Street Journal.

After wrapping up nearly three years as Head of Growth at Uber, he’s joined Andreessen Horowitz as a general partner to help build the next generation of great companies.

I hosted Andrew on our podcast to chat about the changing landscape of customer acquisition, how his “Law of Shitty Clickthroughs” manifests itself in today’s growth channels, and what the rest of us can learn from the likes of Dropbox and Uber. If you enjoy the conversation check out more episodes. You can subscribe on iTunes, stream on Spotify or grab the RSS feed in your player of choice.

What follows is a lightly edited transcript of the conversation. Short on time? Here are five quick takeaways:

  1. Collaborative tools like Dropbox and Slack benefit from built-in virality, where teams adopt them together – and they represent a tidal wave of software products that truly understand the relationships between people.
  2. When your users go through a high-consideration, high-intent signup funnel, like Uber drivers, the key to growth is understanding where folks fall off along the way and finding ways to simplify or shorten that process.
  3. In high-profile cases where growth peaks and crashes, there are often two problems working in concert: The acquisition model might be a single channel, and/or the product might serve an infrequent need, like a mattress or a car. This creates an acquisition treadmill with built-in natural churn.
  4. “The Law of Shitty Clickthroughs” posits that successful channels will become less efficient over time, thanks to a crowding effect that exhausts potential users. Those working in growth and retention must continually seek “fresh powder.”
  5. Growth teams commonly make the mistake of picking random, off-the-shelf KPIs without thinking about how they all fit together. First zero in on a strategy for achieving your desired outcome, and then pick high quality metrics to validate your tests.

Adam Risman: Andrew, welcome to Inside Intercom. You just started your new role at Andreessen Horowitz, and it’s a homecoming for you in that you were in the VC world previously. How are you settling in?

Andrew Chen: I’m wrapping up my fourth week at the firm, and it’s been incredible. The people are really great. It’s such a positive and happy job to have, with some of the best entrepreneurs out there coming to tell you about all the ways they’re going to change the world.

Adam: What drew you back? Was there a particular challenge or an itch you wanted to scratch?

Andrew: Definitely. Figuring out how to grow your business – how you acquire new customers, how you retain them, and how you engage them – is such an important topic for entrepreneurs. I found that after a couple of years at Uber, where I was laser-focused on ride sharing, it really excited me to bring all the knowledge and skills I’ve built over my career to actually help a lot of different entrepreneurs make a big impact across the ecosystem.

Secondly, Andreessen Horowitz is the firm that, for me as an entrepreneur, I’ve always wanted to work with. I’ve known Marc and Ben for a long time, and they originally seed-funded a startup of mine many years ago. It was just such an attractive thing to work somewhere where you have an awesome group of entrepreneurs who are in it to help other entrepreneurs.

Adam: A lot of our listeners are going to know you best through your writing; isn’t that how Marc originally found you back in 2007?

Andrew: That’s right. I moved to the Bay Area about 10 years ago, and I was writing down everything I was learning in my first year. At the time, everyone was like, “Are you crazy? This is your competitive advantage. Why are you writing everything down?” But one of the things that got me excited was saying, “I’m going to give this all away because I’m going to meet really amazing, interesting people.” My first year in the Bay Area, I actually got a cold email from Marc, who was working on his own stuff at the time. It kind of went from there.

What Dropbox can teach us about virality

Adam: You’ve gotten to work with a slew of interesting companies over the years: Gusto, Product Hunt, Angel List, even Boba Guys. You’ve also worked with Dropbox, who just had their very successful IPO. When I think about growth and Dropbox, Drew Houston’s classic talk from the 2010 Startup Lessons Learned Conference immediately comes to mind. He shares the story of how they were spending $200 or $300 to acquire a customer when the product was worth $99, and as a result, they shifted their approach toward virality. How did you get connected with Dropbox, and what can we learn from their story?

Andrew: Drew and Arash Ferdowsi started the company and put it through Y Combinator. I had gotten to know a lot of the folks within the YC community, including Drew. During that period of time he was working with Sean Ellis, who’s a close colleague of mine and coined the term “growth hacking.” We would spend time together and talk about a lot of these interesting challenges.

Dropbox is super unique and innovative today because of this thread they’ve been following over a long period of time, which is to take something that’s just part of your workflow – storing files – and making it spread because of the way people are working with each other. Those early experiments you’re talking about happened during a time when they knew that storing and syncing files had very high retention. Switching to a different service is something that takes a lot of effort.

The interesting early story there is that they had amazing retention but not a lot of top-line growth. The team’s remarkable insight was adding folder sharing. All of a sudden, you’re taking your storage product and then you’re sharing these folders with other people to create built-in, intrinsic virality. I think that’s a missing part of the story: they’re more recognized for the ‘give and get’ disk space, when it fact it’s that intrinsic virality that really powers things. They did an amazing job bringing that all the way up to hundreds of millions of users and then their products for the enterprise, like Paper, are all extensions of that core idea.

Adam: Those products do jobs associated with what Dropbox is built for, and they’re finding ways to grow into those spaces.

Andrew: Right, and that is one of the most exciting parts about products that are happening in the workplace. With B2B, bottoms-up SaaS companies, even Intercom, there is a lot of viral spread because so many people are busy collaborating with each other. Rather than spending years working on a social graph, there’s an interesting workplace graph based on all the people you’re working on projects with and documents you’re editing together. I think that Dropbox, Slack and these other collaborative tools that are emerging are the start of a tidal wave of software products within the enterprise that really understand the relationships between people.

When I’m analyzing the growth strategy of a new product, I skip the homepage.

Adam: Another one of those early learnings from Drew that sticks with me is when he talks about the realization that people weren’t really looking for a way to replace the USB drive in those early days. That seems to be when they changed their strategy.

Andrew: Totally. When I’m analyzing the growth strategy of a new product, I skip the homepage. The homepage is sort of what the company thinks it should be, but people often experience new products through some kind of a side door – like an invite or a shared folder. In the case of YouTube, I very rarely go to the homepage, because most of the time it’s a detail page where a video is playing, and that’s the beginning of your experience. So, when you’re in a world where no one is looking for a shared USB drive, it’s not a compelling pitch. However, if you get an email from a close colleague that says: “Hey, for this critical project we’re working on, here’s a shared folder with all the things that you need to look at. Let’s use this to keep up to date.” Obviously that’s an insanely compelling value proposition and has nothing to do with a shareable USB drive.

Navigating supply and demand at Uber

Adam: Shifting focus from your consulting and advisory roles, you spent the better part of three years in-house at Uber. You joined on the supply side, correct?

Andrew: I started on the driver side of the business, and as everyone knows about marketplaces, the supply side is often the trickiest, hardest side. The reason is very simple: there’s a professionalization that tends to happen. A small number of folks figure out they can make a little money, and then think, “Oh, I might as well make even more money.” These are the eBay power sellers and the folks on Uber who are driving 40-plus hours a week. That group is very finicky, because they’re using the driver app for 10 hours a day. Growing that base is incredibly valuable, so when I joined the company Travis Kalanick and Ed Baker put me on the drivers’ side of the problem, asking: “How do we grow our driver base? How do we acquire more and more folks?” Then, my last year and a half at the company was spent growing the riders’ side. I saw both sides of the marketplace, which was a lot of fun.

Adam: You joined Uber in 2015, so the company and user base were already extremely large. When you have a market that’s so big, where do you start? With established systems already in place, how did you prioritize all the different problems you could have solved?

Andrew: When you look inside any of these hyper-growth companies, what you find – and this is a good signal – is they’ve grown so fast organically they actually haven’t really needed to go super deep on the data, churn models or all the nuances. The first step for anybody coming into one of these teams is to focus on understanding what the hell is going on. The second piece is to then identify some of the key opportunities you want to then execute. Then, you want to measure, iterate and execute that loop as fast as you can.

On the drivers’ side, there were a couple obvious things that needed help. First, anyone who tried to sign up quickly found out that it’s a long process. You have to give a lot of information, you have to give a copy of your driver’s license, and you have to get a background check. In some places, like in Europe, you have to get licensed. So, it can actually take several months to become an Uber driver. This high-consideration, high-intent signup funnel is similar to the problems fintech companies like Wealthfront might face, or a B2B company facing a long, complicated API integration.

A lot of this is really trying to understand the places where folks are falling off. What’s the order of operations in terms of how much you need to ask people? Do you need to ask them for their email? Is a phone number okay? Do you need to actually have their full address up front? Or can you defer that and get them excited about the opportunity before you try to pull them through?

Adam: When you then transitioned to the demand side and concentrated on growing riders, was that a different muscle for you? How did that compare and contrast to the driver side?

Andrew: Drivers are almost like small businesses. They’re very motivated by earnings. They have a long, complicated funnel to get all the way to the end. One example that really works on the supply side is referrals: drivers referring other drivers. Because drivers are in it for earnings, referrals are awesome, and they actually select for drivers that are even better. Now, let’s compare that to the riders’ side, which is usually much simpler because you just put in your phone number and install the app.

Adam: You want them to have that “ah-ha” moment: the car shows up, they get in, and it’s seamless.

Andrew: Exactly. You still need a credit card in many cases, but in other parts of the world Uber goes with cash, so that lowers the friction even more. You’re talking about a different order of magnitude in terms of the complexity of the funnel, right? So, that’s different.

The other thing is that the channels become different. I was just talking about how referrals work so well for drivers because they’re trying to earn more. Think of it this way: if you have a rider who’s in it to get a discount, what kind of rider are they going to be? Probably one who doesn’t spend as much money. So, referrals actually bring slightly lower quality riders. You find a bunch of nuances in there that are very interesting.

One of the obvious observations about Uber these days is that the drivers’ side has more churn than the riders’ side. The riders start by taking rides to the airport, and they think, “Oh, this is pretty cool. I should take it when I’m out and about.” There’s more of a habit, whereas the drivers are always comparing their earnings with Uber to other opportunities like picking up a part-time job.

Why you need a mechanism for free acquisition

Adam: We’ve seen a lot of high-profile startups (particularly in the ecommerce space) raise hundreds of millions of dollars and go all-in on acquisition. Then, they end up crashing back to earth because they don’t have strong retention. Why do we keep seeing this, and what’s the big lesson there?

A natural network forms where every user has the opportunity to acquire one
of their coworkers.

Andrew: This is one of the reasons why B2B SaaS companies have a recurring revenue model. It’s also why a transactional marketplace like Uber, where you have more riders who can actually use it every day for commuting, is nice. That regularity and habit formation means you have better lifetime value. It also means the engagement can power organic acquisition, because you naturally tell your friends about it. Going back to the Dropbox example, or looking at Slack, a natural network forms where every user has the opportunity to acquire one of their coworkers. Another example is DocuSign, where folks who are collaborating within a workflow involve other people from across companies. That’s going to be even more viral than something that only exists within a company. How many folks have discovered Intercom because they saw the little window on the bottom right and thought, “I want that too”? You get all of this free acquisition.

When I look at some of the high-profile cases where it didn’t work, I see a couple of things that work in concert to make it more difficult. First, you have an acquisition model that is a single channel. Maybe it’s Facebook ads, maybe it’s Google ads, maybe it’s SEO – but you don’t have any natural virality. Second, specific to ecommerce, if you’re buying something like a mattress or a car, that happens very infrequently. Because of that, you end up in an acquisition treadmill, where you’ve got to run really, really fast and then – if you’re on a single point of failure on your acquisition channel – there’s an arbitrage for a period of time. If you hit it at exactly the right moment, you can build a pretty decent company. But eventually you should just plan on losing it, right? This is another reason why a lot of gaming companies are hard to fund from a venture perspective: there’s built-in natural churn. Dating apps are also like this. You have that combined with the need to actually buy the traffic because it’s very hard in a dating app to say, “Oh, you should download this too.” That doesn’t make sense.

If you’re building something in fintech or healthcare, these are all things you have to be very careful with and make sure you understand how those dynamics are going to play out long-term.

Fighting channel fatigue

Adam: You wrote a great piece in 2017 outlining an economy where startups are getting cheaper to build but more expensive to grow. Your core thesis was that virality is naturally a channel that is peaking. What should listeners consider as a result of that?

Andrew: The idea is that, especially in pure consumer products, there was a period of time where we had address book importers: you got an invite to a product from a friend, and you were like, “Oh my god, what is this? This is so cool. I want to use this.” And people just got used to that. Eventually, we got to a point, especially now that we’ve gone to mobile, where we don’t have contact importers that work as effectively as the ones before. This is also because email spam and text spam are very different things. There are lots of laws around the latter with the Telephone Consumer Protection Act, and intermediaries like Twilio have a very strict stance on that stuff. What this means is that virality is much harder, and the spammy kind of virality we saw during the Facebook days is not there any more.

So, you have a few options: you could work in a different area where these channels haven’t been exhausted yet. My calendar has all the information about whom I’m meeting on a day-to-day basis. The documents I’m editing and everyone else’s edits on those documents tell me who’s interested in the topics I’m interested in. My email inbox is completely obvious. Even some of the other tools like Slack and Asana give great signals on whom I’m collaborating with. But I’ve actually seen very few products that are built on that idea. It’s this workplace graph that’s just sitting there. So, I’m really excited to see how people take consumer ideas, bring them into the workplace and then adjust them. For instance, in a workplace you don’t need to ‘follow’ your coworkers; you’re on teams automatically, you know you’re on the same email domains, and it’s much easier in many ways.

The other way, within consumer products, is you have to figure out how to make a lot more money and then use different forms of paid acquisition. If you are a product that figures out an awesome consumer subscription business – or you’ve figured out a high-ticket item like housing or cars – all of a sudden you can innovate within paid acquisition. You can do paid referrals or paid ads. You can figure out different kinds of incentives. On a total side tangent, we’re very early on a lot of the crypto applications, but if we fast-forward a couple of years, people are going to play around with a lot of really innovative approaches, whether they’re referrals or a different kind of incentivized engagement.

Adam: Looking at this from a higher level, eventually there will always be diminishing returns on these channels. That’s an idea developed in one of your most famous essays, “The Law of Shitty Clickthroughs.” In the time since you wrote that, how have you seen that observation materialize in new channels that have emerged?

Inevitably whatever worked in the past will no longer work.

Andrew: To summarize the idea, the very first banner ad was for HotWired, and it had a clickthrough rate of more than 70%. Now 20 years later, you look at the average clickthrough rate and it’s like .05%. It’s very low, and anyone who has worked in the industry long enough has seen this happen with email, SMS and all sorts of things for a bunch of reasons. You have competition, and you have the platforms themselves saying, “Hey, we need to clamp down on this.” There’s literally habituation from end users who are thinking, “Oh, it used to be fun to get a invite from my friend, but now I’m getting it all the time.” It’s just less effective, because you have a crowding effect.

The reason why I call it “The Law of Shitty Clickthroughs” is that it’s something that has been with us for a really long time and will continue to be. For all of us in marketing and growth, that means we have to continually find the fresh powder, because inevitably whatever worked in the past will no longer work. By the time a case study has been published on Medium about something that works, it’s probably done. Everyone still has to do it, but then you have to move beyond that.

A lot of the interesting work happening out there ends up on these “frontier platforms.” These are areas where maybe some of the big companies haven’t quite wised up yet; maybe they haven’t started experimenting; maybe the channel is a little too small. These are things like Alexa Skills.

One big area I have found really fascinating is the ecosystem that’s being built around gaming right now. You can livestream things, you can do voice chat, you can do all of these different things around ephemeral networks of players who are getting together over a short period of time to play one game. You’re not going to want to add all these folks to your Skype or Google Hangouts because you are literally just coming together for one game. However, a product that understands that ephemeral network can then build a whole ecosystem around it, and that’s what we’ve seen with Discord and Twitch.

It behooves all of us in the industry to stay on top of these trends and to see what’s working, because otherwise we’re in constant competition where all of our stuff stops working over time.

Unlocking the best insights in growth

Adam: One place where you’ve done an admirable job of trying to communicate those higher ideas is through Reforge with Brian Balfour. You just finished the Retention Series, and you’ve also got the Growth Series. What educational void is the team trying to fill with these programs?

Andrew: Brian Balfour was previously the VP of growth at HubSpot, which invented inbound marketing and a bunch of other important concepts. Brian and I have known each other for a long time. We write the same kind of long-form content, and we tend to be as thoughtful as possible. We try not do the “quick tips and tricks” thing. We really have come to relate on that, and we talk often about how the current form of executive education is kind of broken. It needs to be augmented, because especially in technology, we need to learn frontier skill sets constantly. We need to become lifelong learners, because if you master something, and then two years later there’s a new platform and a whole new ecosystem of startups, you just have to do it over and over again.

Brian and I have started with growth as the first vertical. Brian’s the CEO, I’m on the board, and we basically try to gather folks who are masters of the frontier skill set. We literally ask people,“Hey, who’s the smartest person you know on retention? Who’s the smartest person you know for viral growth on bottoms-up SaaS?” We gather all of those folks and patch them together so you can get real-life interaction with them and pick their brain. Within these frontier skill sets, many of the most amazing practitioners haven’t written their ideas down, because it’s changing all of the time. Brian and the team are capturing all of that.

Andrew’s lightning round

Adam: To close out, we’ve got a lightning round of questions we’ve been asking our growth guests. Short answers are totally fine, but feel free to expand on anything you want. What’s your favorite underused growth tactic?

We forget what a delightful, fun experience and a human voice can do.

Andrew: One of the most important things – especially when it comes to consumer and these days in the work place – is that your product has to be fun. We’ve gotten into a world where we’re so busy measuring and optimizing everything that we forget what a delightful, fun experience and a human voice can do to these response rates.

Adam: What book has most influenced your thinking?

Andrew: I love recommending this book. It’s called “My Life in Advertising,” and it’s the biography of Claude Hopkins, the man who actually invented coupons. He invented stunt marketing back in the day where he’d put things in the middle of malls, like the world’s largest cake. The reason I find it so compelling is that he’s a guy who invented a lot of new channels and strategies that people have built on for many decades since.

Adam: Speaking of people you admire, whom in the growth community do you think we have the most to learn from?

Andrew: First, obviously: Brian Balfour, Casey Winters and Shawn Clowes. Those three guys are involved in Reforge with me for a reason. They’re the most intelligent, thoughtful people from different corners of the growth ecosystem. I also have learned a ton working with Ed Baker and Aaron Schildkrout at Uber, and funny enough, they both started previous companies in the online dating world. Online dating, of course, is a two-sided market that’s hyperlocal, and they had just amazing instincts coming into another two-sided, hyperlocal marketplace with Uber.

Adam: Favorite recent onboarding experience?

Andrew: We’re so used to the world of digital experiences, but the problem with consumers is that when you send a push-notification, you have to compete with all the other notifications, right? One of the most amazing new trends is the way internet-connected physical objects interrupt your real life experience as you’re walking around. One example is all the new LimeBikes that are now in San Francisco, where the onboarding experience is walking around the city and seeing a green thing sitting there – and watching people on their scooters and bikes, having so much fun with big smiles on their faces. That is an amazing onboarding experience. I think as we see more internet-connected devices and products, we’re going to see more of this phenomenon where we figure out how to optimizing things and make them more interesting and presentable.

Adam: You’ve consulted with a lot of growth teams; what’s one common mistake you keep seeing them make when it comes to running experiments?

Andrew: A lot of folks spend their time picking the metrics first and then trying to increase them as much as possible. That’s a good place to start, but the problem is that you have to be so careful about picking your metrics. And in fact, the thing you should pick first is your strategy: “Hey, I’m going to make money on these business customers who are going to pay me, and then I’m going to use that money to buy more business customers.” Or some kind of loop like that. Then, you pick the metrics that validate whether the strategy is working, and you run the experiments afterward. It’s very easy to get caught up in picking random, off-the-shelf KPIs, like MAU or MRR, without really thinking through how it all fits together.

Adam: Where can our listeners go to follow what’s next for you here at Andreessen Horowitz?

Andrew: I am finishing my first month at the firm, and I’m really excited to dedicate a lot more time to writing. Through all of my time at Uber, I maybe wrote half a dozen essays. So, I’m going to try to get into a cadence of posting a couple of times a month cadence on my blog, which is andrewchen.co.

Adam: Careful, we’re going to hold you to it. Andrew, this has been awesome. Thanks so much for inviting us over to your new digs and the coffee and warm hospitality.