Google Photos Product Lead and Bump Cofounder David Lieb with Gustaf Alströmer

Welcome to the podcast!

Guests: Hey, thanks! Thank you so much.

So today we have David Liebe. He is a product director at Google, specifically for Google Photos. What some people might not know is you are also a co-founder of Bump, and Bump was one of the biggest apps on the App Store for the first years of the iPhone. How did that happen?

David: Yeah, totally random! So, I was in business school. I had been an engineer and I kind of went to the dark side and went to business school in the fall of 2008. This was right when the iPhone had come out a year earlier. They just opened up the App Store to third-party developers that summer. So I showed up to business school and I met a couple people in the computer science department who had built apps over the summer. This one guy I met, he had built an app that looked for open Wi-Fi networks. It just told you which ones were open and which ones had like a pay or whatever it's called, a password. He made two hundred thousand dollars that summer selling that. I was like, okay, we should build an app here.

At the same time, I was in business school and I was meeting all my new classmates. We were doing that thing that we still honestly do today, which was, “Hey, tell me your phone number. I'll type it in, then I'll call your phone. You just missed the call.” Then you ask me how to spell my name. One day, I did that ten times and I was like, okay, somebody should solve this. Oh, there's apps now! I could probably solve it.

So I emailed a guy I met. I didn't really work with him directly, but I had met him at my previous job at Texas Instruments. I said, "Hey, any interest in building an iPhone app?" At the time, he was like, "I think Android's going to be bigger." I said, "Let's start with iPhone! I think it'll be better." So we just started hacking away, trying to build this app to make it easier to share your phone number with people you met. It was a total side project. I was like a full-time business school student. He was doing his own thing in California. We just kind of worked on it for a while.

I met my third co-founder, Jake Mintz, at business school. He was also just a random classmate of mine. We met, and we both had worked at Texas Instruments. We started to talk about that, and we brought him on board. The three of us just kind of did the whole thing ourselves in our spare time. We put it on the App Store and people started using it. We were like, "Oh!" This was in 2008. We launched March 27th, 2009.

At the time, it was very non-obvious that apps were going to be big. I remember watching the first like fart apps, you know? Silly, trivial things. A lot of people viewed Bump as a silly, trivial thing too, right? It was like this novelty of the iPhone, like, "Oh yeah, it's got an accelerometer! You can bump your phones together, and it does something!" But that actually was like the engine that made it grow.

We launched it in March. We got a few people started using it. It was growing slowly, but then because of that novelty and the fact that the iPhone was a thing in the news around that time, every reporter wanted to talk about the coolest apps on the iPhone. I remember I met one reporter from the Chicago Tribune at like a random business school conference. I emailed him and I'm like, "Hey, I've got this cool app you might want to write about." He wrote about it.

Then I did this little hack where I took his article and I forwarded it to like the next highest person on the journalism rank. I think I mailed it to David Pogue at the New York Times. I was like, "Hey, the Chicago Tribune has got this already, but you could still write about it!" Then he wrote about it. I took that and I went up the notch. Eventually, it just got pretty popular. I think we were maybe on the like 200 on the top app chart at the time that the billionth app was going to be downloaded from the App Store.

Because we were the top 200 app, we were getting a lot of downloads that day. The probability of us being the billionth download was like 2%, I think, on the day that happened, and it was us. I got a call in Chicago. I was in Chicago. I got a call from the woman who runs all of developer relations from Apple. She's like, "Hey, good morning! You were the billionth app downloaded yesterday! We'd like to tell the world about it, but I wanted to call you to make sure that your servers are gonna be able to handle it." I was like, "Oh yeah, let me call you back."

So I called Andy. I'm like, "Andy, are we gonna be able to do this?" He was like, "I don't know." I said, "Okay, that sounds like a yes." So just we said, "Yeah, go for it!" And that then put us on the map. Like that was the thing that got us globally. Everybody heard of us.

What happened is, yeah, I mean, Apple did a marketing effort around it, but then later, after we joined YC that summer, they put us into an international TV commercial. This was the like, “there's an app for it.” People might remember. We were part of that, and we didn't really know we were gonna be a part of it. The way Apple works is they're very secretive. They say like, "Would you please sign this waiver that says we can use your brand in various things?" I'm like, "Okay, sure!"

One day during YC, we were looking at our graphs and our servers just like—like a factor of a thousand—in ten minutes. We were like, "What's going on?" There must be a bug. Our friends started pinging us saying, "Hey, wait! I just saw you during Dancing with the Stars!" I'm like, "Oh, wow!" That's the point where our servers totally melted down. Everything broke.

This is the first moment where I realized the value of the YC Network. I sent an email out to, you know, YC founders, and I said, "Hey, does anybody know how to—I think we were using a patch, like an Apache web server in our stack at the time. I said, "Is there anybody an expert at this?" They came over to our office and just sat at our desks and they said, "Step away from the keyboards; we'll fix it." They just totally fixed it for us.

That was the first moment where I was like, "Oh, YC is really valuable." I remember when I was talking to Paul Graham about how we should do this, and he said, "I won't be able to explain it a priori why it's really valuable, but I guarantee you'll see that it's very valuable." That was the first moment where I was like, "Yep, he was right."

I'm curious about how you got into YC then—so like, on your duty nobody before housing to be like…

David: Ya know, we didn't know anything about it, honestly. Jake and I were in business school. We were trying to like learn about tech because we, like, everybody in business school, you're supposed to go get an internship for the summer. So we were like, "Oh, where should we get our internships?" So we were like reading TechCrunch a bunch. We kept seeing this thing called Y Combinator. I was like, "What is that?" It turns out we knew a couple people who had gone through it already, and so we went and talked to those people.

I knew Savage, who had done Watt Vision, I think was his first one. I went and talked to him and he was like, "Yeah, you should totally do it!" So we ended up applying and got in. The interview process was fascinating to me as an outsider in this world.

Interviewer: Who was interviewing you?

David: Our interview was PG, Jessica, Robert Morris, and Trevor.

Interviewer: I can only imagine. What were their responses when you're like, "Okay, so this is an AB sherry consecration?"

David: It was fascinating! We walked in and we sat down and literally before I could open my mouth, Paul said, "Give us your phone, we want to try it." I was like, "Okay!" So I handed our phones over to him, and they just started playing around. RTM was like, "I'm gonna try to hack it and break it. Let's all bump at the same time and see if we can break it!" Right?

I do distinctly remember Jessica being like, "So how do you guys know each other? Tell me about that." That was a moment where I was like, "Oh, okay, she's interested in understanding how the founders know each other, what our backgrounds are like, what type of people we are," whereas Trevor and Robert were just like hacking away at the tech, right? Nerds.

There was a fascinating moment. The most surprising sentence that was spoken in our interview was at some point PG just kind of detached and put his head down and was kind of just thinking. He said, "How does this become bigger than Google?" I was like, "Dude, are you crazy?" But that just shows how he thinks that you can take something that seems so frivolous at the beginning and turn it into something potentially that would be really big.

That was so cool. And so where did you establish a metric early on during YC? Like what was your goal before demo day?

David: We were really anchored on how many people used it and how many times they used it. That's like the main thing that we were focused on. In hindsight, that probably was not the right thing to be focused on, but I think especially for us as total newbies to this space, we didn't know anything better.

This was actually one of the big learnings that we had at Bump. We grew like crazy. We got up to 150 million downloads of our app, and I think at the peak we were at like 10 million monthly active users. So it was like, wow, a very big property! Especially ten years ago on the internet, that was really big. But the thing we were looking at was like how many people use it and how many bumps were happening per day and per month.

What we weren't tracking as precisely as we should have was: do the people who use it today keep using it, how frequently do they use it, and what is the retention curve? I remember when we were raising money, all the VCs would ask, "What's your cohort retention curve look like?" I would be like, "Oh, it's good!" Then I would go and Google cohort retention curve to learn what that is.

In hindsight, now I realize it's like the single most important thing. With Bump, the actual dynamic was that our long-term cohort retention was really good actually if you measured it over a wide enough time period. Basically, people kept it on their phone, and they would definitely use it at some point in the future. The problem was the frequency of use was very long.

We thought we were gonna be able to build a business off of a large user base using this product, but if a large user base uses your product infrequently, you have to figure out a way to extract a lot of value each time they use it. For Bump, that value was marginal. It was like, "Yeah, it's nice. It's convenient. I could solve my problem in another way and it's not that hard."

Ultimately, that was the thing that kind of killed it as a business for us.

Interviewer: Hmm, but surely you tried things, right? Like what you try to push into before you ultimately decided to do something with photos?

David: Yeah, we thought about a bunch of ideas like, "Should we just charge for the app or should we have a freemium model where there's a certain feature set that you would get if you would pay?" We tried in-app purchases for different like stickers and things like that. We tested all these things as like tiny 1% tests.

We tried ads in the product also as a little test, and we basically concluded we could get maybe $1 for every user of the product per year on average. Then we realized like, "Oh, we've got 10 million monthly active users; that's a nice business!" But it's not a business that you go raise huge amounts of VC funding to build.

That was the point—this was in like beginning of 2012 maybe—where we realized, "Oh, hmm. We need a better plan," because our plan all to that point was just grow, grow, grow! Facebook did it. Twitter's doing it. We'll just do it too and it'll be great.

I think we didn't understand the fundamental differences between our business and those businesses. What made the product grow? Like what made Bump be so big?

David: It 100% was word-of-mouth distribution. We spent no money on customer acquisition, no money on marketing. I used to tell the story that the amount of money we ever spent on marketing, I think was like $42. It was for me to buy a videotape to put into a borrowed camcorder and a black piece of felt to put behind, so I could make the demo video for Bump in my apartment.

But yeah, it was all word of mouth. People who got the app thought it was cool enough that they wanted to be like, "Hey! Go stuff, try it! I want to show you this cool app I just downloaded." What we thought was gonna be an inhibitor to growth, which is this chicken-and-egg problem—like, you can't use it unless I have it—actually turned out to be the driving force behind our distribution, because it was like novel and cool enough that people were willing to take that investment to say, "Yeah, just download it! I'll wait!"

That worked! You thought about pragmatic fate quite a bit—you mentioned that—and wanted that talks to Y Combinator. So we'll come to that in a second, but there are three different parts you worked in the last couple years.

Interviewer: What's your like, when you think about the word product-market fit for the width in terms of Bump, like how do you analyze that today?

David: Product-market fit, I still think it's a very subjective concept. I don't think you can really measure it, because you can have very great product-market fit with some customer base, but as you try to expand your customer base, you might find, "Oh, yeah! There's not product-market fit for these other people over here." Saying that you've reached product-market fit, I think that is really like it's a staged question.

Ultimately, if you want to be huge in a consumer space, you've got to get to a large group of people. The question is: do you have product-market fit for that large group of people? Even with Google Photos today, like my answer would be—we certainly have product-market fit for a large group of people in the world, but there's another group of people who use our product that I wouldn't say is product-market fit necessarily. There's a whole lot more things that we need to understand and build for that, you know, when you get to like a billion users or more, the next person who joins your product is very different from the first person who joined your product, and so you've got to really understand that.

Interviewer: Got it! So after Bump, you guys tried another product—it just, so, Flock came around. Tell us about that.

David: So we built Bump to share the contact-sharing problem, and people started using it for that. Apple eventually added an API to access photos on the phone and we thought, "Okay, let's add that! Let's see what happens! It's pretty cheap to do." So we added that feature and people started using that.

It turned out that that became the biggest single use case of Bump: sharing photos. One day in, I guess, the beginning of 2012, when I was looking at our dashboards, I was like, "I think it's not gonna keep growing. What are we gonna do?" I thought back to the advice that I heard PG give us back in YC, which was, "When you're stuck, go talk to your users. They will always tell you what they want."

So I said, "Okay guys, give me the list of the top 100 users of Bump in the world. I'm gonna try to go have phone calls with them." So Jake and I that day went into a conference room and just tried to call as many of those people on the phone as we could.

We said, "Hey, we're the founders of Bump. We just wanted to hear what you think of it, why you use it," that sort of thing. What we heard was basically most of those people said, "Oh, I don't use it for contact sharing; I use it to share photos with my husband." And we're like, "Wait, what? You can just email the photos! Why do you do it?"

You know these people would tell us, "Well, it's just easier, and they get the full resolution file, and I don't have to worry about attachments and emails bouncing; like it's just easy." We thought, "Oh." That then led us to: okay, there's clearly a problem in the photo-sharing space amongst your friends or family, but the product we built is like kind of the most hard product to use to solve that problem because both people have to download this app.

You have to stop what you're doing and like both decide, "Okay, we're gonna do this bump thing now," either physically touch. Like when you're building a product, you try to minimize friction of using your product, and with Bump it was almost like we maximized the friction!

So we were like, "Okay, let's try to build a product that minimizes the friction rather than maximizes it." That led us to this product called Flock that we built and launched.

It was really a very convenient turn of events, where at the same time that Jake and I were sitting in that conference room roughly having that learning or that insight, our engineering team was developing the Bump algorithm, which was really like a pattern-matching early type of AI system that would use a bunch of signals to predict whether you bumped with another phone.

Because it was all server-based—there's no right, there's no again FC—and it was all kind of like a magic trick! Our engineers said, "Hey, Dave, I can tell you who's gonna bump with whom tomorrow." I'm like, "How can you do that?" And the answer was they would go look at where and when people took photos on their phone!

So they looked at the metadata of their camera roll and they could say, "I'm pretty sure Gustaf's gonna bump with Craig tomorrow; there's an 80% probability that they do." I'm like, "No way!" What they did is they figured out that today you two were at the YC office and you took a photo, and you took a photo at the same time. And they're like, "Probably you're gonna bump those photos to each other."

That was the key insight on the tech side that allowed us to kind of marry those two insights and create this product, Flock. So Flock looked at your Facebook friends and figured out when you took a photo at the same time as when your other friend was there and then it basically did a suggested sharing prompt to you to say, "Hey, it looks like you might want to share this photo, and just press this one button and we'll share the photo to the other person."

We built that app. We used it ourselves. We used it with our friends and family before we launched, and we realized like, "Wow, this is really great! It just makes my life so much easier; I'm getting all these photos from my friends that I was never gonna get otherwise!"

Then we launched it, and nobody downloaded it! We were like, "Aw, shucks!" This—and what we didn't really understand—which we learned through this experience, was this is a Chris Dixon blog post, which is that like come for the tool, stay for the network idea, which is that the first users of your product have to have some core utility that makes it useful for them even when no one else in the world has the product, and with Flock, it was like, "I download Flock, and it says, 'Great! Go convince your friends to download this app so that you can get some value out of it.'"

Ninety-nine percent of people were like, "Cool, home screen!" and they never downloaded it! So that was the key insight there, and we learned was that if we wanted to solve this automated photo-sharing problem, we had to upstream ourselves and go one step higher in the stack and actually be the camera roll.

If we were the camera roll, then everybody was gonna be using the product anyway, and then all this sharing stuff would actually work. That led us to the third product that we built.

Interviewer: Well, what gives you that insight, though, that you needed to go upstream?

David: Basically, none of our friends were using it, and we asked them about their lives. So just user conversation: I would like talk to you, you know, like a lot of the YC Network. I’d go talk to them. I'm like, "Hey! I see that like I looked in our logs. I see that you downloaded the app but you're not using it! How come?" And people would say, you know, and you have to interpret what they say, but they would say like, "Oh, you know, none of my friends have it yet, so once they get it then it's gonna be awesome!"

And there's something about the physical morality and the physical clumsiness that people see. There are some apps that use – or like Pokémon Go use—we're running around the street like holding up your phone. People are gonna take notice, but that's not really the case with most apps. You just look at your phone and, "Oh, I have no idea what he's doing."

The physical nature of Bump I think made a big difference. I remember the first time I saw someone in the wild like bump that wasn't my friend. It was like a really visceral feeling. I'm like, "Wow! They just used Bump!"

Okay, so I think that was a big part of it, and this was kind of a magical time, a critical type of photo-sharing like 2012. Instagram was like a year in.

Interviewer: Could you anticipate at the time where photo-sharing was going to go?

David: It's very easy in hindsight to always say, "Oh yeah, of course!" I mean, billion, two billion dollar thing. But I like to think that we intellectually understood it, but I don't think we did. I think we more emotionally understood it or personally understood it, and I think we really used this technique that PG talks about, which is if you want to come up with great startup ideas, just try to live at the edge of the future—be the power user of whatever thing you're interested in—and then things will just be obvious to you.

You won't even like intellectually understand them; you'll just be like, "This is useful and cool! I'm gonna go build it!" I think that's what happened to us. Yeah, we were like in hindsight looking back, it's now obvious to me, but like growing up, I was always the person in my family that would like organize the photos and like make sure the photos were taken. I was the person who would always ask my parents, "Hey, can we do a slideshow night?" You'd tell me like who these people in these photos were.

Looking backwards, it's obvious that I was interested in this space. But at the time, we were all using iPhones pretty a lot, and we were taking a bunch of photos. We were seeing these problems of like, "Oh, my iPhone's almost full! I better flush it to my Mac!" We were doing the cable connection, but normal people were just like, "Yeah, my phone's getting full. I'm gonna delete some photos!" We were like, "Oh, that's unfortunate! You shouldn't have to do that!"

I think that kind of led us from Flock. It made us realize there was a whole other set of problems when it comes to photos that nobody was really solving yet. We were personally experiencing them, and so that combined with the failure of Flock and the success of it for people who did get it made us realize we should build this third product.

So the third product we built was this product we called Photo Roll. It was really like a better photo gallery app. This was right on the heels of when Inbox launched. I don't know if people remember this, but it was like a better email client. We were gonna try to pull an Inbox and build a better photo product and just distribute it on the App Store, and people could download it, replace their default app with it.

We built a prototype of it. I was using it on my phone and it was really great. It had Flock built into it on the side, and so that made it so that people could just download this better photo gallery app, and then all of a sudden, as their friends started to get it, it just started to light up for this photo-sharing behavior.

Unfortunately, we had this great insight like kind of at the end of our runway from our twenty million dollars that we had raised. We realized, "Okay, this photo product is a winner if we can figure out a way to like have a long runway and build it, but ooh, we're gonna have to go raise a Series C on Bump to try to do it."

We're not even raising that money to go work on this one hundred and fifty million user product that we've got. It was just a very—like we looked at it and said, "I wouldn't invest in that!"

Interviewer: Okay, let's talk about funders, then. So I mentioned that it also after you guys raised from some of the absolute top VCs in the world, like how does that feel from the outside?

David: Yeah, we were very fortunate! We raised from Sequoia for our Series A. Our bridge round before that was Ron Conway, and then our Series B was Marc Andreessen, and Andreessen Horowitz.

We had the cream of the crop in terms of investors. It felt really great, and I think we were ignorant to the truths of the world about like, were we actually gonna be a really successful company or not? We didn't know, and honestly, I don't think any of the investors knew at the beginning. What we saw was mobile as this huge wave that we all knew was gonna be really big, and Bump was the most popular app on mobile.

So like, and it solves some of these core problems that we all believed would be real. I think that was the investment thesis. That was our kind of like investment thesis for dropping out of business school and doing this: it's working, mobile's gonna be big, we gotta give it a shot.

Interviewer: I think that was the same way. How was the fundraising experience like?

David: It was very easy. I feel guilty saying it, but it was super easy. You know, in YC, I think we got to like either—I think it was either four or eight million users during YC, which at the time was just unheard of. So everybody wanted to invest. We met Sequoia folks and we thought, "Oh, they're pretty good, right?"

Then we raised that round, and then I think it was six months later, Mark emailed me and said, "Hey, I want to invest." I'm like, "Mark, I just raised my Series A! I've got two million dollars left! What are you talking about?" He said, "I think it's gonna be big! I want to invest!"

We took the meeting and I think we delayed it like another six months, but then we raised seventeen million from them. It was again on this thesis that we don't know where this is gonna go, but it's gonna go somewhere very interesting, so let's just plow ahead.

In hindsight, I wish I had raised a little bit less money in that Series B because it would have forced us to have some of these existential questions or discussions a little bit earlier. Like because we had so much runway, we could just say like, "Well, we'll figure it out," and we just kind of kept going for maybe an extra year in there.

Interviewer: So that question comes up with funders a lot that I am talking to. So if they're doing really well at fundraising, and they're like “Well, there's more investors want to give me another three million dollars now should I take it?” With that extra dilution...

David: My advice typically is, "Well, if you have discipline and you can put half of the money in a different bank account and don't touch it for 12 or 18 months, then you should take it!" But most people actually don't have that discipline.

That's true! Yeah, and investors want you to use the money. Like yeah, no investor wants to say, "Oh yeah, I've moved some capital from my LP account into a different bank account, but it's just gonna sit there. Don't worry, it's earning 1% interest." Nobody wants to hear that!

So it's very difficult to do that. What I would—the advice I give entrepreneurs is raise more money if you know how you're gonna use that money today! Like if you have no idea how you're gonna raise it and you're just like, "I want some more money," you should only do that if you already have a very proven business model and there's risks that you want to mitigate.

Right? Like I'm an investor in Flexport. They just raised a big round and I think, I doubt he knows exactly how he's gonna use that money today, but he knows like, "I've got a thing that's working! There are existential risks that I want to mitigate!" And so you raise that round.

But at the early stage, yeah, I'm very skeptical of raising money when you don't know why you need it.

Interviewer: In retrospect, do you think you could have pulled it out of the tailspin and turned a profitable product out of all the work you put in?

David: Oh, I think we totally could have made Bump profitable. I don't—I just didn’t think we could have made it profitable at the scale that, you know, Sequoia or Andreessen Horowitz needs it to be.

Interviewer: Okay, so how do we not raise VC funding? I think it would have been a great business!

David: Yeah, we would have— we would have made a freemium model where like 1% of users bought that for $1 or something like that, and it would have been great, right? But then it's also tricky going and thinking about raising Series C for a product that also might not be profitable at all, right?

Yeah, so we looked at Photo Roll and we were like, "People are gonna love this! It's gonna be big!" If we could get it to be like distributed or part of an operating system like either Apple or Google, we think it would be pretty big.

But then we were like, "And that would cost a lot of money if we were gonna do it because we wanted to store all the photos in the cloud." We did the math and we're like, "Okay, storing all the photos of the world is going to be expensive!" So we kind of decided to do this right.

You've got to do it in a big environment where you can have very long-term horizons. And so when we were getting acquired, we had a number of options, and with Google, it really just clicked in terms of like how we would fit into the mission of the company.

The mission of Google is to organize the world's information and make it universally accessible and useful, and for photos, our mission was to store all your photos, be a home for your photos, make them as useful as possible to you, and then get them to the people that want them—which is like exactly mirroring the Google mission!

So that felt really good. The DNA or the cultures of the companies were very much aligned. At Bump, we were kind of like nerdy, quirky people and we were like into physics and into math and that sort of thing, and at Google, we felt at home.

Interviewer: Compared to, you know, when we would talk to Facebook or talk to Apple, the DNA fit just wasn't quite as good.

David: I think so! That was another reason that we…

Interviewer: What was their lead, you know? The Android team? It was her photos team do you join? It was Picasa back in the days, right?

David: Yeah, Google has been on the photos trajectory for a while, beginning with Picasa. There was that acquisition, I think it was in 2005, of Picasa. Then Picasa turned into Picasa Web, and then when Google+ launched, it got rolled into Google+.

So we showed up in the fall of 2013, and Google was full speed ahead on Google+ and they were starting to build more and more photo-related features into Google+ with the goal of getting you to share them on the social network. Our conversation with Google began with the Android team, and it was in the context of like the gallery app on Android could be a lot better and smarter, and here's an app that could do it.

So the acquisition was really to take Photo Roll and turn it into something that would be useful at the Google scale. It was really convenient. We joined Google and we like started to ask around about who's working on what, and we actually were now moved over as part of this photos team inside of Google+.

I showed up and I'm like, "Hey, I've got this idea that I think is gonna work around building a photo gallery app, like a private photo management tool that allows you to share and have this cool AI stuff in it." I was kind of like, but I don't have any AI! Like we're just here. We showed up and we looked around and we were like, "Oh wow! They've got like this huge team working on face recognition, face grouping technology! They've got this other huge team working on understanding the content of images and being able to search images! They've got this other team around scalingly backing up photos at like a huge scale."

I just saw this and I'm like, "Oh! This is it! This is gonna be good!" It's a perfect fit, and what I think we brought to the table as Photo Roll was this insight around like how do people think about their photos? What is the right design of a product? How do you build these sorts of features in a way that would fit into their lives really well?

So the combination of those two things turned into Google Photos and the definition of actually uploading every single photo for every album is actually there right now! That's what you do?

David: Yeah, we do. Yeah, yeah! We upload a lot of photos every day.

Interviewer: Well, what was the insight that led you to create—I understand the sharing between people, the facial-recognition, that’s awesome! Searching is so cool! What was the insight that led you to start doing the animations and stuff like that?

David: So a lot of that stuff is actually already being done inside of Google+. They did it to try to give you something interesting that would be post-worthy, that would make you want to post it on G+. The process that we took when we were designing Google Photos was to use this technique that I like to use a lot, which is pretend that there's a human being doing the thing that your product is going to do. And pretend that human being like doesn't have to sleep, is really smart, has access to all the computing power and brain power in the world—what could that human being do for you?

We were talking about this this morning, but for photos, I thought, "Okay, if my co-founder Andy was my photo assistant and all he did all day long was help me with my photos, what would he do?" We just started brainstorming like what were the things he would do that would be useful to me? Things were like, make sure that he backed up a copy of every photo. So if I lost my phone, I didn't lose my photos.

Cool! He would look at all the photos and he would know who's important to me in my life—he'd know who's my mom, who's my dad, who's my sister, and who's my girlfriend—and he would like write on the back of each of those photos metaphorically, like, "This one contains mom and dad! This one contains Jenna!"

Then we were like, "Well, what else could he do?" Then I'd say, "Oh! I kind of asked him to go learn like Final Cut Pro, and I'd want him to make me cool montage movies of my life!" When there's something interesting, I'd want him to like remind me when it's like the one-year anniversary or something cool that happened in my life or something meaningful. I'd want him to like go edit my photos if I have a photo that's like too dark or messed up, I'd want him to edit it!

If I have photos that are like blurry or crappy, I'd want him to suggest that I get rid of those! We just made this laundry list of things that would actually be really useful to me, but no one was doing because we didn't have that Andy doing it for me.

That was the insight that kind of led to how we designed Google Photos, which was it's your home for your photos, but then it's basically like this really helpful assistant that would just do all this other stuff for you, and that persists to this day and it's working very well, that model.

Interviewer: What's your day-to-day today like? When you come into work at Google, how does it look to run one of the 17 billion billion-dollar user apps in the world?

David: It's great! It's very diverse! How I spend my time is very diverse! I would say I'll answer it on a monthly basis rather than a daily basis just because every day is very different! On a monthly basis, I think I spend, you know, a quarter of my time maybe thinking about the strategy aspects—like where should we take the team? Where should we take the product? What is changing about the world or our user base that we need to understand and kind of move in one direction or another?

Maybe another quarter is like working with our team and helping our team, whether that be like developing managers on our team, figuring out how to structure the team, which leaders we should put in charge of which things—that's another quarter.

Another quarter, which is probably my most pleasurable quarter, is like actually working with people on our team directly to think about what we should do with the product—like thinking about, "Okay, how much should we change for our sharing model?" Like, how can we make it better? Or—you know, one that I'm working on right now is this like photo assistant, the ability that we have, like what's the next step? How do we like do something more with that?

That's another quarter. And then the last quarter is like stuff that you've got to do if you're running a team, and it's, you know, help people with their feedback. How do they get better as a team? How do they get better as an individual? Kind of the stuff that you don't think about when you're a startup with ten people, but when you're a company with hundreds of people, you're like, "Okay, I need to come up with a good system where we can give feedback to people on our team, and we can help develop them, and we can make them better."

Interviewer: So this was the next thing I wanted to talk to you about—like how do you think about your work now? Coming from, you know, basically three people to a small startup to now a very big team—like, obviously, management strategies are different, how you spend your time is different, but like, how do you actually feel about it?

David: It's very different! People ask me this question all the time. They're like, "What was it like? What was the culture shock like going from a startup to a big company?" I think I might be lucky, but it doesn't feel like that at all to me. It feels like our startup acquired a multi-hundred person team. It really does!

I think it just speaks volumes to the fact that our DNA match with Google was really good, and that when we joined, we eventually were given the autonomy and authority to actually like chart the path for this product. And that was a rocky road, but like eventually we got there.

So yeah, I think that's how it feels to me, and it's really a luxury! It feels great! After kind of having that insight at Google, when I invest in companies, this is a question I explicitly ask them: if you could go acquire a big company with the purpose of furthering your vision, which big company would you acquire? Hearing the answer to that question helps me understand a) their ambition and b) how do they think about where they want to take this thing long term?

So I often ask this question and they're like, "Wait! We don't have any money! We can't acquire a company! What are you talking about?" I'm like, "No, no! It's a thought experiment!" Just yeah, and I think it's really fascinating to hear those answers.

So one thing I'm curious about—so I would argue that Google Photos is like one of the magical AI experiences, where like you actually take a great product and you add machine learning and AI and make it magical. There are lots of startups that come to YC and apply with ideas that relate to AI, but how do you think about—like how would you advise them to say, "I have this product and I have this cool machine learning capability I'm adding to it?"

Interviewer: What do you typically, when you see those companies, what do you ask them? What do you advise them?

David: Yeah, I think the thing that really worked about Google Photos is it was one of the first examples of AI solving real problems that people could resonate with. A lot of the AI stuff I see does not do that, and you're like, "How does the AI actually uniquely make this better?"

So in thinking about how you could apply AI, I kind of think about it in two different vectors. One vector is: is there a thing that human beings can do either with skill or time or resources that they don't do at scale today, but we could bring a computer in to do that job? That's kind of like what Google Photos—that's our approach in many ways.

I think there's a lot of problems that could be solved in the world that today you could solve easily with a good human being, but you just don't have that many good human beings at the right price!

Alright, so that's one vector that I love to see, and that one I fully believe in! The other vector is problems that human beings cannot solve today—things like how do you optimize the power usage of a data center? Like a human being could never solve that problem! It's beyond our brains!

That's the other vector where AI can do really fascinating things, and I think that one is a lot harder to know a priori whether it's going to work or not. In many cases, there are much bigger payoffs if you can get it right!

So that's how I think about it. My advice to people who are like an AI startup is: tell me which of those two are you? Like, are you trying to solve this unsolvable human problem? Are you just trying to scale a very solvable human problem to a whole lot of people or at a much cheaper price?

And when you're doing that, what are the things you're gonna go build? Like, what is the solution that you're gonna try to show to your users? If your answer is something that the solution—even though the AI tech is super cool—if the solution is like, "Yeah, that's nice, but"—whatever—then it's not gonna work, right?

It seems like this attention to the user and attention to product is like a core belief of yours. It's not about the technology, right?

David: And this was a thing that I noticed at Google when I joined. I think since that period, like in the five years that have passed, Google has really made a transformation to focus much more on the user. But at the time there was still a lot of like cool tech being developed, and everybody at Google is a technologist, and they're like, "This is really cool!"

But they weren't asking the question as much about, "Well, how does it solve a user problem?" And is it really a core user problem that we should solve? I think, you know, as Sundar has taken over at Google, he's really tried to focus the company on the user and how can we be helpful to users, and I think that has just really played out in the form of Google Photos.

Interviewer: Is it different to talk to users at scale when you've built users versus when you're going to a startup? How is that different?

David: I've got a tab open in my Chrome browser for each of our platforms for iOS, Android, and web—that is the raw stream of feedback that you can send in the app. Like every day we get like thousands of reports, and so I can't even read them all anymore!

Whereas at the startup, I can read every single email. In fact, at Bump, until I think through the acquisition, I was the only customer support person! Like I wrote back to every single email, and I blocked like an hour or an hour and a half every day doing it. It was kind of one of these things that can't scale, and it was probably a bad use of my time at some point.

But it taught me that like you listen to people and generally they will tell you what you should do. You just need to understand how to interpret what they're saying. It's a great practice! I think that's something we should recommend all early startups to do because it gives you just the recollection connection with the users.

One thing we did at Bump, which I really enjoyed, is we would take our team and go to bars and do user testing and talk to people. One reason we did it at bars is that people are in a more social environment, so like bumping was something that you would actually do there.

Whereas like managing your photos, maybe it's harder to find the moment when people are in that mode. But the other thing is when people have a couple drinks, they like actually tell you what they think!

Whereas when you bring someone into a lab and set them down behind the mirrored glass, they know they’re being studied, and so they behave differently. This is a thing that I think is challenging at scale: how do you do that? You can't just like tell Google employees like, "Yeah, go to bars and ask people what they think!" It's not gonna work!

So I would say it's an area that in general could probably be improved. How do you create these learnings and this empathy with users at scale for a team that's large and for a user base that's very large?

Interviewer: Do you have like a system that’s just parsing the text and looking for keywords?

David: Yeah, we have some AI systems that read the feedback, try to cluster it into different clusters, and then some humans look at that and try to understand if there are interesting insights. We've got a user research team that goes out into the world and asks people.

We do trips! Like a couple summers ago, I spent two weeks in India just living in India trying to understand what about Indian people is different than what is about, you know, people in Mountain View.

Yeah, there's a lot of differences! So we try to do stuff like that. We bring people into our office also, but again, like that has its own challenges in terms of—I mean, knowing that you're being tested—but yeah, we try to do as much of this as possible, and I think—I honestly—I put a lot of weight in what intuition you can build by just talking to your users.

I think we overlook the fact that we've got the best computing platform that has ever been built, and it's our brains! We can use it to give it inputs and then let it do its magic and pattern match, and then output our intuition.

I think that that is really what your gut is, and it really is like a sophisticated machine learning system that we should actually utilize. So I'm a big believer in building what you think is right.

Interviewer: Absolutely! So a question from Twitter related to products, Lamby Day Aka Mallafa (hopefully I got that right) asks: What did David focus on too much that he thinks was a mistake now when you were a rookie product lead?

David: Oh, great question! What did we focus on too much? One thing at Bump that we focused on a lot was like how good the system was—like how performant was it? Measuring, you know, we obsessed over like microseconds here and milliseconds there.

I love that culture, and it's a culture that we try to really push on Google Photos as well. But I think if you focus on that a little too much early on, you might be missing the much bigger picture, which is like even if you made it super awesome, people aren't gonna care.

I think that that is an insight that is hard to really have when you're in the moment in the details. Your system is running, and you're like, "Crap! It's like really slow right now! All the users are complaining! It's so easy to just be like, 'Great, let's go fix that!' And it's very hard to pull back and say, 'Maybe that's not the problem we should be fixing. Maybe there's some other problem that we need to understand!'"

I think that is the trap that a lot of companies fall into. This is the—this is the Sam Altman—it's like one of my favorite quotes from Sam is, I think it's like he says, "The meta problem that kills most startups is working on the wrong thing." It's not like screwing up whatever you're working on; it's not your competitors; it's that you were just focused on the wrong thing.

That is probably my number one advice.

Interviewer: And so now that you're working on it at large scale—like these product inside—what degree of buy-in do you have to feel before you ship it out to a billion people?

David: Great question! We do a number of things to get confidence in some product that we're gonna build or some feature that we're gonna build. The first thing we do is use it ourselves, okay?

And if we think it's not working, then it’s very likely it's not going to work in the market! This is a luxury that we have because we build Google Photos; it's a product that we're building for ourselves and for a lot of other people who are not like us, but we are at least somewhat representative of the user base compared to, you know, if we were building a product for like, you know, kids that are under eight years old. Like I personally don't have a ton of intuition about that, and I wouldn't be able to know innately whether this is gonna work or not, so that's a whole lot more challenging problem.

I have so much respect for people who can solve problems that are where they are not the user. For us, we are the user! So the first step is: do we like it? Often, you know, I challenge our team where we've done experiments, we've run stuff, we were about to launch something, and I just asked the team, "What do you guys think about it? Do you like it? Do you use it? Like, what's your most annoying part of this product?"

Often they're like, "Well, I'm not the target user, but this is kind of annoying to me," and I'm like, "We should get that right—come on!"

So that's number one. Number two is we do a bunch of testing with users before we actually launch. Even in an experiment, we make prototypes; we bring it out to people and you can learn a lot of usability things like whether the actual flows will work, but you can't really test: do they care by doing those sorts of tests.

We also do experiments. We roll things out to 1%. We've got like a— I won't talk about it, but we've got a pretty big product in experiment right now that we're excited about, and we're watching how people are using it. We're trying to keep it on the down-low.

That’s another thing that we do, and we learn a lot from that. I think the combination of all these things gives us confidence that, yep, this is a thing that we at least have high enough confidence that we should roll it out and see what happens.

Interviewer: Hmm, yeah!

David: So, Laminae Akumal Alpha is asking you: did you make any early mistakes when you joined Google?

Interviewer: So, like frankly, like what kind of is new large organizations to you—like what mistakes?

David: Ooh! Certainly they would say yes! I think, let's see. There are two problems and they're wrapped together. The meta problem that I made—a meta mistake I made was I really believed in this vision for what we could build in the photo space so much that I was just determined to solve it.

I think the skill that I needed to develop inside of a big company—which is very different than what I had at the startup—was I now had like three or four bosses above me that I needed to convince. At the startup, I just had to like have a board meeting and say, "Hey, Mark, I'm gonna do this!" and he's like, "Cool, do it!"

Whereas here, it was like, "Hey, I gotta convince this person and then I've got to use their support to garner support from this other group of people and then I've got to take that and create support from the CEO," right?

It's a very different process! I was not very adept at doing that and I just kind of, I used the modality of being the startup founder and just saying, "We need to do this! Come on, everybody! Let's go!" and I'm gonna convince you!

Whereas at Google, I needed to be much more delicate! I don't think I did a great job of that. One of the sub-components of that was: how can you do that at a big company? I have learned now is you can really get a lot of support at the grassroots level.

So when I ran into these problems going up the stack at Google, I kind of retrenched and said, "Great! Let me go talk to the people on the ground, actually talking to customers, actually building the product as opposed to the people managing big teams doing it."

When I would go talk to those people, a lot of them said like, "Oh yeah, I get it! Let's do it! This makes a lot of sense!" By building some support there, we could then go in mass to the leader and say, "Hey, we all think it's right! Here are all the perspectives. Here are all the diverse opinions: this is what we think we should do."

Interviewer: In five or ten years, do you see yourself working for a large company or startup?

David: That's a great question! I love startups! I love the rawness of startups. The purity there—like every problem you face is there are no layers of abstraction; you're just in it and you're seeing the true problem. I love that!

It's a thing I try to keep doing at Google. The challenge, I guess, is when I look at like what I could go start next or whether I want to join a startup or start any startup, I kind of have the luxury of thinking about it in two dimensions: one is what reach will I have? How many people in the world can I affect with this thing I want to build?

The other dimension is to what degree can I affect their lives in a positive way? Looking at Google Photos, it’s like we have a path to affect billions of users—the majority of people on the planet—and we have a path to get there; that's pretty amazing! Not many products can do that!

And then on the importance or depth of the experience, I believe there's not a lot of things more valuable to you than the record of your life—your memories. Maybe the only things on top of that are like being alive—like your health and maybe connections in the moment to the people around you—like, otherwise it's one of the most essential parts of the human experience.

So I look at that and I'm like, "I think this is probably the most impactful thing I can be doing right now."

Interviewer: I actually give that advice when the companies ask me, "Hey! I got this position offer from this large company." I always try to put like all the financial stuff aside and everything else aside, and the main question to me is always: can you execute any vision towards a much larger audience in a better way inside this company or not?

If you really believe that being part of this other large company is the way for you to get to your users and execute a vision, then that is the most important question to answer.

David: Yeah! And a corollary is if you don't have that alignment and you do go to that company, it's going to be a disaster!

Interviewer: Yeah! Basically most acquisitions don't turn out well, and it's fascinating! You make this big decision for your company that we're gonna go get acquired by blank, but then if you look at like how many minutes or hours did I spend with the people I will be spending my entire work life with?

David: It’s usually really, really small. Like laughably small! Most of the discussions are on the financial exam! It’s not actually about the product outcome. That should be where most—out of the challenges I faced when I got to Google were misalignments of vision around what Google should do in the photo space.

I wish we had had those conversations beforehand, the problem is the conversations need to be with these like senior level people at Google, and they don't have time to spend like an entire weekend like spitballing with me about some random startup entrepreneur like the future of what we should do, right?

That might or might not getting quite right!

It's this like asymmetric problem! Now that I'm on the other side at Google, I try to empathize with my former self, and like, you know, we've done acquisitions that photos, and I try to just spend a lot of time with people, make sure that we're aligned on like what we want to build, because inevitably when you join, something's gonna change and you're gonna have to tweak your plan in some way! If you don't have that fundamental alignment, it's gonna be a disaster!

Interviewer: Hmm, so that's a thing I think is really important!

You said something really interesting about humans and their attachment to photos—so I wanted to kind of touch on this at some point.

David: Yes! So have you learned any kind of larger truths about humanity by paying attention to all of this stuff all the time for years?

Interviewer: Probably, I don't know if I'm great at articulating them! Let's see…

David: Yeah! I think one probably the number one truth I've seen is the power of nostalgia. I think we all feel it and we all know that feeling, but seeing it from the perspective that I get to see it, it's one of the most powerful things in the human experience, I think!

We've tapped into it in a variety of little ways, but I think there's a much larger thing that we're still trying to go learn and figure out. But yeah, I think we each think about important moments in our lives, and they could be important from like a grandiose perspective or just like little moments in our lives that stick with us.

Usually in that moment, there's like an image that you have in your brain or like a song that you associate with that moment or like a smell. There's something that like triggers you—and people love thinking about the past! It's just an innate thing!

I think it's an evolutionary thing that those of us who had that trait are better at understanding where we've been and better at surviving in the future, so I think it is baked into us as humans, but it is a really powerful phenomenon!

Interviewer: Have you mmm everything in Fast and Slow?

David: Oh, yeah! Talks about so like how important photographic memories are for happiness!

Interviewer: Yeah, I haven't— I totally believe it. One of our goals with Google Photos is to basically give everyone a photographic memory. The challenge is our brains aren't like most of us at least don't have this like perfectly linear evenly distributed photographic memory!

The way our brains work is we pick important moments and really like write a lot to the disk during those moments, and then we kind of don’t worry about the rest! One of the challenges we have with Google Photos is we now have all the photos that you've taken in your life!

How do we decide which ones are the important ones that we should like really focus on and try to help you relive and engage with? Which ones are like not important even though they may seem important to our algorithms?

And then on the other side, which ones are like not memories that you want to re-engage with? This is a big challenge and it's a very difficult problem, but we all have moments in our lives or people in our lives that were part of our life before, and it was an experience we had, but going forward, it’s not a thing that we want to dwell on, and that's a really difficult challenge.

I'm like, I would say we're getting it okay, but not great, and I want to—that's an area I want to improve.

Interviewer: Yeah, so I have one final question! As a product lead—I was—we had similar roles at Airbnb—I spent a lot of time looking at other products and new products. What are your top like other new apps? The European iPhoto space? Alright, in general!

David: Like, actually just—you feel now like these are important to me?

Interviewer: Consumer products?

David: Yeah! Or just apps generally!

Interviewer: Well, in terms of like categories that I see that I'm interested in spanning like everything—one category that I think is fascinating is basically taking like mundane systems that humans have created and replacing them with smart technology.

Flexport is one example of that. I just invested last YC batch in this company Canary Technologies—building software for hotels. Like seems so mundane and like who cares about that! But it's a big problem! A lot of people go to hotels, and we need to make the experiences better! One way to do it is to make the software better! So that's like one category that I'm very interested in.

Another category that I'm interested in is taking things that we all take for granted, like eating food or having healthcare, and applying some of these technologies or things to make those things better. So one I just invested in also in the last batch is Seattle Food Tech. They're making a meatless chicken nugget!

I just ate them the other day—the newest version of it—and they're delicious, and it's like obvious to me that, oh yeah, we're gonna... this will be a thing! Some form of this will be a thing! So applying technology to these like very mundane physical parts of our lives is another category that's interesting.

On the consumer-social yada-yada side, it's really tough! Like, I see lots of cool stuff where I'm like, “There's something cool in there! But this thing's gonna fail!” I think that all the time, and I try to invest in them because some fraction of them will become really cool!

It's a different world today! Maybe we were like, oh, it definitely is—seven years ago!

Interviewer: Yeah! We've like—developers have gone and like figured out all the obvious wins already in the consumer space, and the ones that emerge are the ones that seem like totally off the cut!

David: Right! Like, who would have known ten years ago that Snapchat would have been like a popular thing? Or that the stories format would have been a popular thing? It's very hard to predict until you build it and see it, and then you're like, "Oh yeah, that's no!"

Interviewer: Well, let alone a lip syncing product because, you know, one thing on the consumer side that I've come to understand is there is a difference between something that is really popular for a while and something that is a durable essential human need.

Through the Instagram phase, given where I am in photos, I got to see all the Instagram competitors where they have like some new cool format, right? Like it turns your photos into art or it makes this part of the photo move, or it’s this cool looping effect.

All of them have the thesis: like we're gonna take this and turn it into Instagram! My feedback was always like, but Instagram already exists, so you're not solving a durable human need!

Interviewer: Look at the timing or what's the specific thing that they did that made them succeed?

David: I think had those things happened before Instagram happened, they could have pulled the same thing off. I mean, Instagram pulled the same playbook, right?

Yeah! So, alright!

Interviewer: Yes! Thank you so much for coming in!

David: Thank you so much! Yes, great!