field notes

Field Notes: Marketplace Tactics with Deb Liu (Part 1)

Connie Chan

Posted April 12, 2023
ABOUT FIELD NOTES This is Field Notes, a new video podcast series by a16z that explores the business models and behaviors that are changing consumer technology. Subscribe to the a16z channel on YouTube so you don't miss an episode. In this episode, host Connie Chan talks to Deb Liu, the CEO of Ancestry and the former VP of App Commerce at Meta. The conversation covers all things marketplaces: what metrics to focus on, how to balance growth and focus, and how to turn user behavior into business strategy.

Episode Transcript

Connie Chan: Deb, let’s start with Ancestry, where you’re currently the CEO. I heard that you joined this company remotely—you hadn’t met them in person. We know a lot of people now, even today, are doing interviews on Zoom. What tips do you have for people thinking about interviewing remotely, especially executives? Tell me about joining a company as CEO without ever having stepped into the building.

Deb Liu: Well, it was really intimidating. They were picking a CEO for a company that they had just bought for $4.7 billion. Blackstone [the investment firm that acquired Ancestry in 2020] did the entire interview process on Zoom. It was such an honor to meet so many people and to be able to connect with them multiple times, but at the same time, you really don’t know what someone’s like unless you really meet them [in person].

The other part that was really funny was when I was deciding whether to join they set me up with the executives. I met them all in one day. And instead of telling them who I was, they set up the meetings blind on their side. So instead of telling them who I was, they just said “You’re meeting the new CEO,” and then I showed up on Zoom. And I didn’t know! The first couple people I met, I didn’t realize that they hadn’t been told who told who I was in advance.

So some of them just kind of looked at me and I said, “Are you okay? Do you wanna Google me?” And each of them said yes. And then I realized what they had done. And it was so interesting. To walk in blind and meet your new manager or your new boss with no information about her was jarring for them, too.

Connie: What an interesting introduction into your role as CEO. I want to talk about Ancestry, specifically, because I think a lot of people don’t realize that Ancestry has done a lot of pioneering work with AI. Recently, you deployed AI handwriting recognition to search the U.S. census. In what other ways is Ancestry using AI? What are you excited about?

Deb: Well, AI is going to revolutionize our entire industry. I mean, we previously [digitized] the 1940 Census. It took us nine months and hundreds of people hand-keying and indexing the records. Think about that: there are like a million pages of records. And so what they did was they just dropped the pages [online]—it’s just images. And so, the 1940 Census took us nine months and we spent a ton of money and a ton of time making it happen. 

So for 1950, we wanted to be absolutely prepared. So we built an AI model. We actually created pages, we tested it, we tried all these different things. And then it took only nine days to digitize the entire census.

On the other side, I think AI will really transform storytelling, too. You know, because someone’s life story is now available on Ancestry. We can actually build stories about people’s lives based on just the records people have saved there. And so I just think that that’s going to help people really bring the stories of their families to life: Not just connecting over records of birth/marriage/death, but also the life that they actually led. 

Connie: You guys have a massive data set on people’s heritages—their grandparents, their great-grandparents, their great-great-grandparents. How are you thinking about the data opportunity inside Ancestry?

Deb: Well, first we have 40 billion records, 75% of which is completely exclusive to Ancestry. And so these are records like draft cards and birth certificates, but also we digitize a lot of land records and local records, as well. And so this past year we had a record year: 5 billion pieces of content digitized. And then this coming year, for 2023, we’re going to triple that because of the AI work that we’re doing. We’ll be bringing 15 billion more records [online]. 

Connie: When I think about AI, I often wonder if we fast forward five years from now, what could be possible? Is there anything you can share about—a couple years from now, five years from now, 10 years—what will we be able to know about our past through a company like Ancestry?

Deb: Well, part of it is really the storytelling. Not just the “what”—we have a lot of those records—but why. Can we find letters? Can we digitize people’s records of the letters that they have, the photos they have? Can we actually record the oral history of our families and then transcribe them so that we can help bring that storytelling to life? Can we automatically create books around the story of your family without you having to do hundreds and hundreds of hours of writing? These are the most powerful things that I think we could bring to life, and I hope that that will come to pass soon.

Connie: So storytelling actually has been part of an arc in your career, because you also spent a lot of time figuring out how to bridge relationships at Facebook. You spent a good chunk of your career there leading platform, payments, games, and marketplaces. What are some of the parallels that you saw between the missions of the two companies or the type of work you were doing? 

Deb: Well, part of the work that I did at Facebook all those years was really about connection. You know, the theme for Facebook marketplace when we started it was connecting people through commerce. And you know, it’s not really just about transactions. It’s about actually changing people’s lives through how you connect them. And so the same thing at Ancestry is really connecting people to their own family story. And social media and Facebook is about storytelling—it’s just you’re telling it in the present, and Ancestry is telling the story in the past.

Those things will come together, because today becomes the past very, very soon. And that’s why this job attracted me, because a lot of the work that I did at Facebook was about connecting people in the present with transactions, connecting with your community, connecting through commerce. And now, it’s another way of connecting. And that’s the theme of my career, which has been technology can really bring us closer together if we do it right.

Connie: You’ve definitely spent your time at companies working on projects that are bringing people together. And I want to get a little bit more tactical, because I know a lot of our listeners are founders who are building products. You’ve built a lot of those things at Meta almost from the ground up. So let’s start with marketplaces: What are some types of advice that you give to marketplace founders? It may be they’re doing a cold start, or maybe they’re trying to figure out how to scale, or they’re hitting issues where supply is greater than demand or vice versa.

Deb: You know, I think the biggest thing about building a marketplace is: what is your right to win? What is the category and what is the narrow thing that you have an advantage in that no one else has, whether that’s an interest, whether that’s sellers or buyers, whether that’s a niche or brand. Knowing who you are and what it is that makes you special is the first and most important step to building a marketplace.

Connie: Does that mean, though, if there’s already a marketplace that addresses those buyers and sellers that there’s no room for a new competitor?

Deb: No, you can have a totally different take on it. I mean, look at eBay. I worked at eBay for many years, and what eBay did really well was all of the categories. But what they did poorly was any individual category. And you’ve seen entire categories get moved off of eBay because they couldn’t do really well by, say, rental properties. They couldn’t do really well by collectibles. They couldn’t do really well by handmade sellers.

So you can see these marketplaces saying, “Hey, if we just focus on these buyers and sellers, we could do this better. And we don’t have all the categories, but we have the ones that really matter to our customers.”

Connie: But then, those marketplaces where they pick something off, eventually the goal is to recreate that large, biggest player that covers a lot of categories, right? It seems like every company that picks off something from Craigslist, the ultimate goal is to build back Craigslist.

Deb: But if you think about it, for example, Airbnb. Airbnb grew out of Craigslist with local rentals. But they haven’t ended up in every category, right?

Connie: That’s right.

Deb: So focus is incredibly important. And the same thing is true if you look at Etsy. Like, they’re just not going to be in used cars and car parts and things like that. And so [it’s about] really knowing who you’re going to be and what you’re going to be and having a vision for: What is the thing that I want to win first? 

And then look at the concentric circles. Some of those concentric circles will never overlap. Etsy will probably never sell car parts. 

Connie: How do you figure out, though, what is the next concentric circle to tackle? Because Etsy started with handmade things, but now there’s a bunch of stuff on there that’s not handmade, actually. Lots of cross-border stuff. People say a lot of the Chinese cross-border Amazon sellers are now all on Etsy. Whether that’s true or not, I’m not sure, but the point being: If you are building a marketplace, how do you figure out what is the next category to go after and when to do it?

Deb: First, what is the most similar category to what you’re doing? So Amazon did books because there’s ISBN—it was very structured, very easy to look up, categories were already there. But the next thing they did was CDs, because music published by large publishers had a very similar categorization system. You know, one of the most recent things they’ve been trying to do for a long time is soft goods—basically, clothing. Which is very, very hard when you have a very structured system. And so really where you start will take you longer and longer to get out to the edges.

I think the thing about Etsy is, what they did was they attracted a type of buyer and now they’re trying to figure out what categories meet that type of buyer. I am an Etsy buyer. I actually buy these alpaca fur bears that my daughter collects on there. And you know, a lot of them are made in Peru and imported into America. And they’re handmade by artisans, but those are not the sellers. The sellers are actually third parties. My daughter collects them and they’re beautiful. But as a buyer of handmade goods, this category is like an extension of something that I would buy, because I’ve bought a number of other handmade things. And so the question has to become: How true do you stay to your core? And then: In which next concentric circle are you willing to go? And really meeting your buyers where they are. What are the things that they’re looking for? 

But I do think they have to be really careful because it can feel like it has the inauthenticity of manufactured goods. You know, these bears are still handmade, but they’re just not made by the seller. That’s one thing. But then if they’re manufactured in mass production, they’re gonna feel really inauthentic to me as a buyer. And so I think they’re walking a really fine line of growth as well as authenticity. 

Connie: And so as you’ve studied all these large existing marketplaces—and you’ve built some of them and worked at some of them yourself—what are the big things you think they did right that people might not be aware of and some missteps that you think might have caused them to go off track?

Deb: Well, I think what eBay did that was really special was having people trust each other online to send money in the mail. That just seemed so crazy at the time. But I remember I was an early eBay seller and people would just mail me money orders. It’s like mailing cash in the mail to another person. People would send me checks, which had all their information on it. But it was that trust system that they built, the reputation system was something where: You can send money to someone else, and they’re gonna send you a thing in the mail. That just seems so crazy, and yet their reputation system was able to build that connection between people. I think PayPal augmented that, there was a layer so you didn’t feel like you had to research trust. So that actually accelerated the growth. I think what changed for eBay was that the reputation system actually became a double-edged sword.

Part of it was that every employee of eBay knew that you didn’t want to buy from somebody that had a 4.8 star or less. Because in order to give someone negative feedback as a seller, you basically had to burn your buyer account. You had to be willing to never come back, because the seller was going to do something called retaliatory negative feedback, which was to give you negative feedback as a buyer.

And so the system was set up to protect sellers and to hurt buyers in some negative way, which wasn’t known to the company for a long time. And part of that really hurt the reputation. Because now somebody who’s a 4.6 seller—which you would think is very good in the world—suddenly became a really negative experience. And yet people didn’t know how to complain. They would complain quietly to PayPal. They would get their money back. But then they would never actually publish a negative star because they were afraid to burn their account. And so you’re creating incentives on both sides. And it’s important how you think about those incentives. So I wrote an article really thinking about ratings and reviews in detail.

On a lot of sites, it’s a mutual interaction. So on some sites, like Airbnb, for example, you both give each other feedback, but that feedback is revealed at the same time. That way, you can’t see the feedback before you write your review. There’s no retaliation. 

Another way to do it: eBay just doesn’t allow buyers to get negative feedback from sellers. Because buyers completed their part of the transaction by paying and by making the purchase. And so they changed their policy at some point. But the point is really thinking hard about the reputation and trust system can make or break you over time.

Connie: What about a marketplace like Amazon? What are some things that you think that they’ve done particularly well?

Deb: Well, what they did wonderfully was that when they didn’t have the inventory, they supplemented it. They were actually a first-party seller, and then they tested auctions and it didn’t work. And then suddenly they said, you know what? We’re going to have this “Buy Box,” and a third-party marketplace and just see what happens. And suddenly it opened up a whole new avenue to bring in a long tail of product that they could never have had and warehoused in their own inventory. And so it really brought selection to the fore and it allowed people to compete for the Buy Box.

And so you’re competing on quality of service, plus you’re competing on price. And so they use both dimensions. Basically, you’ll have no sales if you’re not in the Buy Box. And so what they did was they gamified two sellers. 

And so I think each marketplace really needs to think about: What are you incenting? Are you incenting good behavior, or just trying to prevent bad behavior? And those things matter a lot over time. 

Connie: And how early should founders be thinking about this? Is this core to how the whole product is set up from day one?

Deb: It should be absolutely core to how the product is set up. So one of the biggest challenges you see in marketplaces is: there’s seller fraud, there’s buyer fraud, and there’s collusion fraud. And every marketplace needs to think about all of those things at the same time. If you don’t build that in from the start, you could really lose the thread and it could cost you a lot of money.

Connie: Yeah, that makes sense. What advice do you give to people who are in the growth stage? Where they actually have a great product, there’s good liquidity, but they’re not hitting the growth targets that they’re looking for. How do you counsel marketplaces on growth?

Deb: Well, I think sometimes it’s easy to say: Let’s just open new categories. Right? The question is, are you actually deepening your relationship every single day with your buyers? And so really looking hard at the retention curves, the J curves, and seeing if you’re retaining them. And is your product getting better every day? For Facebook Marketplaceour first metric wasn’t transactions—because we couldn’t measure them—it was how often people were coming [to the site]. It was weekly to monthly actives. Because as a classifieds marketplace, it’s about frequency. People don’t buy anything unless they come. And they don’t come if you’re not top of mind. And so really it was about: How do we get them there? And then once we got them there, can we get them to message people? And then once we get them to message people, can we get sellers to mark things as sold?

And so it was really a single step. But then over time, we actually saw traffic going up, so people came back more often. If they made a transaction satisfactorily, they would come back more often over time. And so then we would get the J curves. And I think a lot of marketplaces are like: Let’s just look at raw tonnage of transactions.

Connie: Yeah. Most marketplaces focus on metrics that are all numbers. It’s all dollars. It’s not the number of users or frequency of visits or any of that.

Deb: But if you actually search “leaky bucket” and eBay, there was an article written about this. At some point you become a leaky bucket, and then you burn through entire countries of buyers. And that becomes a huge problem. And I think they’re trying to fix that now. But I think one of the big issues that marketplaces have is: You can always bring more people through the door with enough dollars. The question is, can you deepen your relationship with them? What is your actual 30 day, 90 day, 180 day retention?

Connie: What are some things that you did at Facebook Marketplace to get buyers to spend more time, check it more frequently, or have better retention? Anything you can share there?

Deb: Well, in a marketplace where the goods are completely dynamic…So, for example, you’re unlikely to go to Airbnb if you’re not traveling, right? You just don’t go and browse and see what’s available. But in a browse marketplace, it’s really important to bring people in. So we gave people reasons to come in. Every time you loaded the marketplace tab, the top of the tab was completely different. It had different items on it, so it never felt static. Then we added this thing called “top picks.”

Connie: So even if there was nothing new, you would shuffle it?

Deb: Yeah. We would just shuffle it. So we were constantly reshuffling because we wanted people to have a feeling of: there’s something new here every day. And so really knowing what you occupy and where you are matters a lot to actually how you build the experience.

Connie: And in that browse scenario—because a lot of marketplaces tend to be a bit more browsing—how do you figure out how to balance the familiar versus the new? You know, maybe there’s sellers that they’ve already purchased from before, or maybe there’s a restaurant they always order on DoorDash. How do you surface that versus showing the net new restaurants that they haven’t seen before?

Deb: Well, I think the best thing is: really understand people’s behaviors. There is a group of people who really only buy one thing. When I go to DoorDash, for example: I have this favorite Mexican restaurant that is across the street from PayPal. And when we order from there, we order like 10 burritos and eat it for, like, several days. And so, they can show someone like me that.

But then when I’m in a different location, they should suggest every single thing that’s possible out there. When we travel, I want to discover new things. I love Vietnamese food, so my kids are like, let’s find the local Vietnamese food and eat banh mi. And so, it should show me a selection of things that I’ve looked for before when I’m in a different place. And so really thinking hard about: what is the job to be done for this person at the moment that they come, and how can I actually take all the information I have on their previous visits and bring it all together? 

Connie:  I really love that idea of focusing not just on the dollars—the top line dollars, the GMV that’s going through, the take rates.  I mean, that all matters. But to your point, if the buyers aren’t browsing, you have fewer and fewer chances to get them to convert to a transaction. So even just figuring out everyday traffic is a really important metric that marketplaces should take into account.

Deb: It’s traffic, but it’s conversion on traffic. But the other thing you said was, it’s not just traffic versus GMV. Are you thinking about GMV or are you thinking about transactions?

And so there was a time, and this was back when Flipkart was growing like crazy, that they really chased GMV. And so they sold a ton of mobile phones because it was a high GMV item. Amazon, on the other hand, was like: how many transactions? I’d rather you buy a $2 tube of toothpaste here than something worth $200. It was all about frequency of use, because now you’re really dependent on it.

Connie: And that drives things like Prime, which unlocks a bunch of other stuff.

Deb: Right. And so the whole point is, their approach was really different. If you’re chasing GMV, you want to sell high-GMV growth categories. If you’re chasing transactions—which is: people come back and buy $2 tooth of toothpaste—then your behavior and what you present to people is really different.

Connie: This is amazing feedback, I think, for a lot of founders to focus on: general traffic and retention, not just GMV. Very, very important.

I also want to talk about a different aspect that I think makes you an amazing role model for a lot of folks in tech, which is that you’re also an author. You are also really big on mentorship. You’re the creator of not only a book, but also a newsletter that a lot of readers—especially females—are consuming to learn how to be successful in tech. Talk a little bit about how you moonlight as a career coach. You’ve mentored hundreds of people. What kind of career advice do you give those lucky mentees of yours? And what’s something tactical that you can share with our audience?

Deb: Well, you know, recently I wrote this article: Tell them what you want. And I know it sounds really crazy, but so many people call me to do these 15 minute coaching sessions. And it’s usually strangers—they text me, I call them. 

And someone’s like, “Well, I want to get promoted, but I don’t know how.” I’m like, “Well, what happened when you told your manager you want to get promoted?” And it’s…silence. And I say, how do you expect to get promoted if you don’t tell somebody? And I realize that sometimes we get in our own way. We know what we’re supposed to do.

And my job is not to tell people what to do, but to unlock in them what they need to know. Because a lot of these people know they should have done that before calling me. And I’m like, why don’t you call me back after you talk to your manager? But the point is, I think sometimes we just assume that things are supposed to work a certain way, because it’s comfortable for us.

But I tell people: If you don’t hear “no” enough, you’re probably not asking enough. 

Connie: How often should we be hearing no?

Deb: Well, I read a lot of stories about founders. They’re like, “I went to 30 VCs.” But you just need one or two people to say “yes.”And the whole point is, if hearing “no” is something that you’re allergic to, you’re never going to be able to raise money. How do you get your first partner? You’re going to be the underdog for a long time. And the answer might be: not now. And then you say: “Thank you. I know this isn’t right for you now. Let’s talk in six months.”

But I think sometimes we get in our own heads. We worry so much about what other people think about us. Instead, treat it as a numbers game. I remember talking to one founder and she just had a list of all the people who had turned her down. But there was one who said yes. I said, how did you deal with that much rejection? And she said, “I realized that my goal was not to get everyone to say yes, but to get the right one to say yes.”

Subscribe to the a16z channel on YouTube so you don’t miss an episode.

More About This Podcast

Field Notes is a video podcast series by a16z general partner Connie Chan on the business models and emerging behaviors driving consumer technology. In a series of candid interviews, Connie talks to the builders, creators, and investors behind the tech that shapes our daily lives.

Learn More