Startup executive and investor Andrew Chen discusses his book "The Cold Start Problem: How to Start and Scale Network Effects", revealing what makes winning networks thrive, why some startups fail to successfully scale, and, most crucially, why products that create and compete using the network effect are vitally important today.

Although software has become easier to build, launching and scaling new products and services remains difficult. Startups face daunting challenges entering the technology ecosystem, including stiff competition, copycats, and ineffective marketing channels. Teams launching new products must consider the advantages of “the network effect,” where a product or service’s value increases as more users engage with it. Apple, Google, Microsoft, and other tech giants utilize network effects, and most tech products incorporate them, whether they're messaging apps, workplace collaboration tools, or marketplaces. Network effects provide a path for fledgling products to break through, attracting new users through viral growth and word of mouth. Yet most entrepreneurs lack the vocabulary and context to describe them—much less understand the fundamental principles that drive the effect. What exactly are network effects? How do teams create and build them into their products? How do products compete in a market where every player has them?

Andrew Chen is a general partner at Andreessen Horowitz where he invests in consumer technology, including social, marketplace, entertainment and gaming experiences. Today, Andrew serves on the boards of All Day Kitchens, Clubhouse, Envoy, Hipcamp, Maven, SandboxVR, Singularity6, Sleeper, Snackpass and Substack. He is a prolific writer and leading voice on mobile, metrics and user growth. He is also a board member and instructor at Reforge, which offers selective growth-focused programs for experienced professionals in marketing, product, data and engineering.

Get the book here: https://goo.gle/3F5CklE.

Moderated by Ali Norup.

Hi, Welcome to this Talks at Google virtual event.

And as the head of Americas for our Google Cloud for Startups program for VC Relationships, I'm very excited today to introduce you to today's guest, Andrew Chen.

Andrew is a general partner at Andreessen Horowitz, where he invests in consumer technology, including social, marketplace, entertainment, and gaming experiences.

Today, Andrew serves on the boards of Allday Kitchens, Clubhouse, Envoy, SnackPass, and Substack.

Andrew is a prolific writer and a leading voice of mobile, metrics, and user growth.

He's also the author of The Cold Start Problem, a book exploring how new startups are launched.

The Cold Start Problem is now available wherever books are sold.

Lastly, a quick disclosure for you that none of the following should be taken as investment advice from Andrew.

And we're going to be able to talk to him about a number of things today.

So with that, Andrew, it is my pleasure to welcome you to Talks at Google.

And it's really a pleasure to be on here and to talk about the new book.

It's actually the day after launch day, and also my birthday.

So we have all these great things happening on the same day.

And again, really, really appreciative of having, of being on here.

And so what I wanted to do is actually just to take a couple of minutes.

And I have just a couple of short slides to just talk about, introduce the book and the concepts in the book.

And then we can just kind of riff from there, if that works for you.

So this book, just to give a little bit of the genesis, it's a book that I've been writing now for actually 3 years, which is really exciting.

And the book is called The Cold Start Problem.

And the subtitle is how to start and scale network effects.

And the reason that I started writing this was just that, in the in the portfolio of Andreessen Horowitz, and we've been lucky to work with some of the best companies and technology, we just found over and over again that at their core, the products that just seem to just completely kick ass and just like grow huge and are just awesome, impactful products are often the ones that are just connecting.

At their core, they connect people for different reasons.

But it's really about connecting people that drives them.

And so Airbnb, you're connecting guests and hosts.

For Figma and for GitHub, you're connecting coworkers and developers and people to work on projects together for products like Clubhouse and Instagram and Pinterest and some of these other companies that we've invested in.

They're really connecting people, connecting content creators as well as consumers.

And so really at the core of all of this, it's really just to extrapolate and to really sort of generalize, 1 of the core driving forces for software eating the world and for technology being as impactful and as successful as it is, is this term that we use to describe these kinds of products, which is they have network effects.

And of course, literally, the product we're using right now for this presentation.

A lot of the fantastic Google products, like Gmail and G Suite and everything, have network effects.

And so anyways, we'll talk all about that.

But really, it's something that makes all these products really powerful.

And so you might ask, well, what are network effects?

And so I'm just going to go through a really simple definition, which is built on the invention of the telephone.

So the phone is kind of a magical thing because when it was first invented, of course, by Alexander Graham Bell and then commercialized by a company called the American Telephone and Telegraph Company, so A-T and T, A-T and T, over

The chairman of the company, Theodore Vail, actually had this amazing quote in it.

And it said, and he said, a telephone without a connection at the other end of the line is not even a toy or a scientific instrument.

It is 1 of the most useless things in the world.

Its value depends on the connection with the other telephone and increases with the number of connections.

And to me, this is a great quote, because if you were to replace a telephone with workplace chat app or telephone with a dating app or something like that, you would basically get a sense of what it's about, which is that you can build an app.

But if you don't actually have the right people in the product at the same time with you, then it's just not going to be useful at all.

And so you're just not going to use a product like like some of the products that I've been discussing, unless other people are also using it in the same way.

And so just to run through a couple examples of this, if you take a product like Dropbox, this is Dropbox paper, and if you are using it, if you open it up at work and maybe use it a bunch for your personal life, but you open it up at work and you notice that none of your co-workers are on it, or you open up Asana, or you open up Slack, or you open up 1 of these other products and none of your co-workers are on it, then you're much, much less likely to use that product.

You'll probably try and figure out what else, What does the team actually use?

Oh, okay, well, they use G-Drive, they use Notion, they use Airtable.

It's like, okay, well, I'll go use those instead because that's what everyone else is using.

And same for a product like Uber, where I worked during some of the hyper growth years and ran a bunch of the growth team.

And the thing that we were the most worried about is when somebody opened the app, and there was no cars available.

That was just so, so bad, because it meant that the network was broken.

And same for a marketplace company like eBay, or same as many products out there.

If you open up Tinder and there's not enough people around you that you are the kind of people you want to date, you're not going to use that app.

And so, so really, this is kind of the core focus of the book.

This is the cold start problem, which is the other side of the coin.

It's basically, yes, when you when you build 1 of these products and you hear about how, whether it's Instagram or it's Discord or it's Slack or any of these products, when they have a lot of users, if the products become more useful and they're gonna grow even faster, But on the other side of the coin, what it means is that if you don't have any users, the product isn't valuable, it's not gonna work, and that'll kill your new startup or your new product that you've been working on at work.

And so the book is really meant to try and talk through that whole lifecycle and how to solve it.

And so I have this five-part framework that I present in the book, which is that all of these companies, all of these new products, start with the cold start problem.

And to solve it, there's all these different techniques.

I talk about using atomic networks and forming atomic networks.

I talk about the easy side of a network and the hard side of a network and where to focus.

And then all of a sudden you should get into the tipping point, which is when you're able to create these networks kind of on demand and you're able to start building the second and the third and the fifth and the tenth 1 and then off you go and then escape velocity where you start to really professionalize your growth marketing approaches and really amplify your network effects and then inevitably as all these products become really, really successful, there's various ceilings that you hit.

Going back to Dropbox for a second, there's a point where that company started, and I was an advisor for many years off and on, and Drew Houston is a friend and he's a fantastic guy.

And 1 of the things they saw was they got to a point where people were using Dropbox as a way to share pirated movies in the Philippines.

And you're like, maybe that's not a good use of the product.

And if all of your money is going towards that, It's not going to be useful for other things.

And then there's always, of course, the issues that are happening in social media, where you have a lot of trolls, you have spammers, you have just this overcrowding.

If you go to YouTube, there's just too many videos.

Or a chat product like Slack, It's an amazing product when you have 50 people on it or 100 people.

But I was using Slack in a 20,000-person organization.

It starts getting really hard because it's very, very overcrowded.

And so you need to start to really build features to solve that.

And then finally, in the framework, I talk about the moat, which is how do you find ultimately build the defensibility in your product such that you can compete against new startups and new entrants into the market.

But that's kind of the main sort of topic of the book.

And it was fantastic, really fun to write.

I was able to interview hundreds of people from all the top tech companies.

I also locked myself in a little closet of an office for a week to read the audio book, even though everyone told me you shouldn't do it.

I figured that it would be, you know, I'm just, this is, I think this book will be a 1 and done.

And so it's great to, I was like, I'm just going to read it for real on there.

But yeah, that's sort of the gist of the book.

We had a series of questions we were going to go through to potentially have you do it.

And I may even have you come and pull up a slide with it as well.

I'm going to take a step back for a second.

We kicked off and we talked about how you're a partner at A16Z.

But when you moved to the Bay, you were an entrepreneur in residence, and you started with writing a blog post and a series of them.

You found writing as a way to create your own network, to some of the parts of the Cold Star Problem, and you definitely saw the additional effects.

What was it specifically that made it so that you were prompted to write this book?

It's something that I was the kind of teenager that kept journals and wrote about things that were happening in my life all the time.

And when I got to the Bay Area, I absolutely had such an amazing time meeting people because I was writing down all the ideas that I was learning.

And so some of the concepts like viral loops and growth hacking and thinking about kind of this intersection between user psychology and metrics and growth was sort of my favorite area to write in.

And so I've written almost 1,000 essays on this topic.

What I decided at 1 point is I just needed to bring it all together and try to answer this question that's kind of been in the back of my head for a long time, which is just, what is it that's so special about some of the quote unquote Silicon Valley?

Of course, none of these companies are actually built in Silicon Valley anymore.

But like these quote unquote Silicon Valley companies, what is so special about Silicon Valley companies that has caused them to make such a big impact in the world?

And the more time that I spent thinking about that, the more I was like, you know, I think I'm going to be able to unify this into 1 core idea that these products are all about building networks.

And that is the thing that has been really special about what's happened with them.

And so yeah, so I wrote the book to collect all of these theories together in 1 place.

And yeah, and the result has been the cold start problem.

So 1 of the things that you say at the beginning of the book is sort of like the network of networks.

You reference Metcalfe's law of the network effect.

If you were to sort of dial it up for us, What are some of the lessons that for all of the founders of startups, the people in the ecosystem that are listening, that you would love to share?

I know there's the framework of the 5 pieces, but if you were to go a step further.

I think There's so much lore in the startup world.

You hear about how Snapchat was started in high schools and Tinder was started in colleges.

And you hear about these stories like Slack and everything.

What is it that all these guys were able to do in the beginning that made their product work and off the ground?

And so I make several arguments in the cold start problem around this.

I think the first argument is it's not an accident that all of these startups begin in these niches.

And they begin in these niches because ultimately there's this idea of an atomic network, which is how many people need to be connected at the same time doing the same thing in order for that product to be valuable.

And it turns out for a product like Zoom, you just need 2 or 3 people.

And you can actually just keep talking and talking and talking, and it's great.

And there are products that are like Slack that are more valuable when 5 to 10 people on the same team are using it.

But then there's products like Airbnb and Uber, where in a city, you actually need hundreds of drivers and need hundreds of listings in order for the product to be valuable.

And so So the first thing is knowing and or having at least a theory on what that threshold actually looks like, I think is actually really, really important.

And so a very good example of this is Slack in the early days, if you actually go back to the story.

I mean, first of all, the whole thing is amazing.

Like Stuart Butterfield, after creating Flickr, actually he created a video game first and then like stopped and then was like, what am I doing?

And then he and Katerina Faye and a bunch of other people all started Flickr together, that was sold to Yahoo, invented photo tagging and a lot of the constructs that we think about today.

The second part of it was, so he started a new game called Glitch, and I encourage everyone to search on YouTube for glitch trailer because it's so amazing.

You're these characters and you're chatting with each other, you're learning about underwater basket weaving, the levels are built on the heads of all of these giants.

And after raising tens of millions of dollars, hiring dozens of people, it just completely bombed in the market.

And he did this wonderful thing of actually continuing on, of continuing on.

And he brought the team down all the way down to, I think it was 6 or 7 people.

1 of the people that, a different Allie actually, that I interviewed for the book, Allie Rail, who's amazing, who's now runs all the product at Slack these days.

And she was 1 of the first 7 people that rebuilt the whole thing.

And the way that he rebuilt it was after, first of all, they actually had a lot of IRC-based chat tools that they were using already.

And then, so it didn't require IRC, so it was like easy for non-technical people.

They made it so that you could kind of, a new company could configure it.

And they sort of knew that because they were a team of just a couple people, that if the chat product was working for them, so meaning that team was an atomic network, and they were able to get 1 of their friends' companies to also onboard, and another set of people to onboard, then they should be able to then build a second, and third, and fourth, and fifth, and tenth atomic network.

And then they could basically create this wait list and and then onboard teams at a time.

And I think that's a very important approach as opposed to just announcing on TechCrunch or whatever and then like a bunch of random people show up and then it's like, oh, none of my co-workers use this.

And then they're just going to quit using the app.

So I think just embedded within that lesson is you start small.

And the reason why you start small is so that you can build that density of the network.

I think it's also the reason why larger companies historically have been so bad at this.

You know, there's a lot of companies that, you know, over the years have tried to compete with, you know, TikTok and with, you know, actually right now, 1 of my companies, Clubhouse, has a whole bunch of, you know, missiles pointed at them.

And historically, companies have not been as good, larger companies have not been as good at this because they basically just take their huge audience and they basically say, hey, you should all try this new shiny little app.

And the problem is that new app doesn't have time to actually coalesce and those networks to build organically bit by bit by bit in order for it to work.

And instead, you get a lot of users, but you get a lot of users who are not connected with each other that aren't talking to each other.

1 of my favorite interviews that I had in the book is, so I was an advisor to Tinder back in the day, back when Sean Rad was still running the company.

1 of the ideas that he talked about, which I thought was just fantastic, is that You have to make the app really, really amazing for the most attractive people in the online dating market.

Can you, yes, please talk about the Tinder case study.

So every network product has what I describe as an easy side and a hard side of

And the easy side of a network is like a YouTube viewer, or it's like a buyer on eBay, or if you want to use a new marketplace, like Whatnot or something like that.

And they just do less work and their livelihood does not depend on it.

If you ride Uber, you just hit a button, you ride for 5 minutes, it is what it is.

But when you are on the other side of the network, That means you're a content creator, you're like a meeting organizer, you're like a seller.

What ends up happening is like your livelihood is like very wrapped up in your use of the product.

You're extremely picky, you are very, very concerned about, you know, how much you're being paid, you have all these like different lessons.

And so, and they're just much more valuable.

And so what happens is in the case of Uber, for instance, a driver, Uber might pay

to buy a driver versus to buy a rider using online ads or referrals or something, I'd only pay 20 bucks.

So there's like a 10X difference between the value of the 2 sides.

Okay, we're gonna come back to Tinder now.

So on Tinder, the most interesting thing on Tinder is that historically online dating has been very focused on just getting people to make profiles.

And then if you go back to the 90s, like the original match.com or eHarmony was just like, you're just trying

And the problem with that is if you are an attractive member of the audience, and frankly, it's skewed more towards women as well as the hard side of the network, what ended up happening was they were just inundated with messages left and right.

Because unlike when you're at a dinner party or at a bar or something like that, you can't, like if 1 person's talking to like 2 or 3 people, you're gonna like talk to their friend.

And so if you were an attractive member of the network, you would get thousands of messages, literally thousands of messages.

And so what ended up happening with Tinder that I thought Sean just did such an amazing job describing is that we all think of the swiping as kind of like this fun thing.

But more importantly, it lets the attractive people in the desirable members of the Tinder network, it actually lets them modulate how fast they're getting matches.

can match with 10 people, and they can stop swiping, and they can just manage their 10 conversations or their 3 conversations or whatever.

If they want to have 500 matches a week, they can do that too.

And, and so I think, and I, so I think 1 of the really key lessons to me in this is that if you understand the, the easy side and the hard side of your network, whatever it may be, that you need to provide kind of this additional functionality for this highly, highly valuable subset of your business, whether that's making something that makes drivers happier, or if you're building workplace tools, something that makes your software developer, like leads happier.

You really have to figure out something there to unlock it.

Because that's been and so so so, you know, just popping back up to the kind of the high level, these concepts, like the hard side and the easy side of the market, the atomic network, thinking about, you know, come for the tools, stay for the network, like all these different playbooks of strategies.

My goal in writing them and collecting them into this book, and the book is effectively 30 case studies on all these different ideas, is to sort of demystify all these stories that would otherwise kind of seem random.

Like then you kind of look at them and they just kind of feel like, oh yeah, it's just kind of lucky that, you know, that happened.

And that was why versus being like, oh wow, okay.

So Uber going from city to city to city, Tinder going from college to college to college, Slack going from team to team to team, actually have a really interesting underlying thematic strategy for why that actually works.

And in the same way, that's why if you're a huge company, just blasting everything to PR and putting links on your homepage and doing all that, well, that doesn't let you build highly dense network to highly dense network and to go strength to strength.

And so as a result, like we've seen that much less effective.

And so, yeah, so the cold start problem is meant to sort of unpack a lot of these historical patterns and ideas into a cohesive framework.

You've done a couple of things there, right?

You're demystifying some of these stories so that we can all sort of take the learnings from them.

I think the other thing that you do that's phenomenal is you show just how disciplined these teams are, right?

So let's go back to Tinder for a second, your network of networks, they start with a small party on a campus.

But they understand, and we'll get to LinkedIn in a second because I love this as a corollary example, they understand that you have to have that quality of the atomic network right at the beginning.

So we'll do early and then mid and late stage.

Well, so Tinder is, I do love that story also because it's such an iconic product that, and it's just so fun what they've been able to do.

And so Tinder in the early years, when it was a product that actually.

Basically, the first version of it had everything, all the elements that you would think of as Tinder.

They just like nailed the product idea from day 1.

And sometimes that happens, you know, the first the first version of Instagram just like nailed it.

You know, first version of Tinder just nailed it.

And so, John Bedean, who's their iOS engineer, had a had a deck of playing cards that he would kind of fiddle with while he was coding.

And 1 day, the very, very, very first version actually just had an X and a check mark.

And that's how you would kind of advance.

But then he was like, oh, you know, swiping's fun.

And like, so he added it in and then boom, off you go with this iconic thing.

So they built basically all the right features.

The problem with building all the right features is that then they went and they started to invite their friends.

And they went on their address books and they started texting their friends.

Again, this is at a point where online dating kind of is a little bit of a stigma.

So imagine this, you're texting your friends and you're kind of like, you're almost saying like, hey, you seem kind of lonely.

You know, it's kind of a little bit like an insult.

And then importantly, I think the Tinder atomic network is not 2 or 3 people.

It's also not thousands of people, but it's probably hundreds of people because you can actually swipe through.

If you think about how many swipes you can do each day, you can actually swipe a couple dozen people a day pretty fast, pretty easily.

And so as a result, you actually need hundreds of profiles in order to make it interesting.

That was their cold star problem was, How do they get enough people in it at once?

And it just turns out, even though they had all the right features, that even when they told their friends about it, it was not working.

And I think that that is such an important thing, just to stop there for 1 second, which is that sometimes you actually have all the right features, you just don't have the right network.

See all Talks at Google transcripts on Youtube

Andrew Chen | The Cold Start Problem: How to Start and Scale Network Effects | Talks at Google