00:00:00 · Sandy Diao, Guillaume Moubeche
[00:00:00] S0: When I say data inspired, I mean that we're looking for directional insights. We're not looking for precision. AI has kind of trained us to expect value really, really quickly and really, really fast. Power law outcomes drive the results, meaning that one or two channels drive the majority of users. I think by definition in the world of AI, there is going to be very few growth moats that remain the same. [00:00:20] S1: Today on BILLIONS, I'm sitting down with someone who believes most growth teams will be obsolete in five years, and she's been right about markets before. Sandy Diao scaled products to 200,000,000 users by leading Pinterest's growth effort as employee number 30. Then joined Descript as their first growth hire, built an affiliate program that drove 25% of all new users, almost entirely self-service. The channel specialist is dead. The winners will be full stack operators who can own the entire funnel, because in a world where anyone can generate content, trust is the only channel that compounds. Sandy, welcome to BILLIONS. [00:01:00] S0: So excited to be here. [00:01:03] S1: I was really, like, impressed by your profile, obviously, and excited, you know, like, to to have this conversation. When you when you joined Pinterest, at first, I think you you wanted to, you know, be very involved in growth, but ended up doing, like, a lot of customer support in the in the early days. What were you trying to look for when doing this kind of customer support interview and also answering tickets?
00:01:29 · Sandy Diao
[00:01:29] S0: Yeah. Well, actually, thinking back to my early days at Pinterest, I don't think I planned for my role at Pinterest to be this full stack growth role, but it just accidentally ended up being that way. And I think the accident was by design. So basically, what I mean by that is I actually joined the company with this role of doing anything possible to help the company grow. And the company's mandate at the time was that our customer support inquiries were on fire, essentially. So, you know, I jumped into essentially the trenches and got behind the computer, you know, answer hundreds of support tickets on a weekly basis. But something got me thinking, which was that every time a customer wrote in, it was jam packed with insights around a... What a customer was happy with, what they had issues with. And I realized that I could actually pull data points around some of these top inquiries. So it was through that that we realized that there were a lot of experiences that were broken. One of the notable ones being that for a lot of new users, how to use Pinterest effectively was something that they needed more guidance on. And so one of the first, you know, early product growth growth marketing initiatives that I got to work on was rewriting our onboarding emails from scratch, which is how you effectively activate and onboard new users. And that insight only came through because we kept repeatedly hearing people have these questions about using the Pinterest product in ways that they weren't supposed to be using it or that there were better ways to use for. So, you know, that became a great insight. And that naturally evolved into all these other experiments and opportunities. So it was really fun to be able to do that, and I think that accidentally evolved into this generalist growth role where I touched product surfaces, ran marketing campaigns, ran scaled sales outbound for the sales team, you know, segmenting the market for consumers, prosumers, mid market, enterprise, etcetera, and so on and so forth. So, you know, kind of thinking back to that origin here, I I think the tickets was actually a great way to actually have customer insights be the starting place for a number of growth experiments afterwards.
00:03:26 · Guillaume Moubeche
[00:03:26] S1: Yeah. That's that's really cool. And you mentioned that you leverage emails to increase activation on specific features. How exactly do you set up everything and track everything? Did you work with a specific data team, or how exactly did that work?
00:03:43 · Sandy Diao
[00:03:43] S0: Yeah. In the early days of using emails for activation, one of the challenges that we had was that we had essentially this benchmark of peer companies. So I think at the time, considered itself to be a social media app, although it is a little bit different if you've, you know, used the product yourself. But I think the challenge is that when you see other companies out there like Twitter, Facebook, and and others set up their emails, you kinda see they have this formula. Right? And it was... At the time, onboarding emails were very education oriented. You know, it was like, you're gonna receive five to 10 emails from us every single day, and here are the things that you should be doing. It was very much treated as early user education, and companies all had that approach. Over time, we realized that, one, the more emails you sent, the more email fatigue there was, and people just stopped opening them. So by the seventh email, no one read what it is that you wanted to sell them. And secondarily, when you actually measured outcomes and engagement based on the emails you're sending, we found that driving specific product actions was far more useful than trying to drive awareness about every possible thing. And instead, the product itself should do the work of educating while the users in the product experience. So we went from this education based approach through our emails to essentially this more data driven action triggered approach. So for example, if a user came in to the platform, they hadn't followed any boards yet, then our our goal would be to get them to follow boards that indicated their interest. If they signed up and they created boards, but they hadn't pinned anything to it yet, then our goal would be to get them to take, you know, a pinning action. So it really was catered and customized based on what a user did or didn't do. So these drop off or fall off states. And a very data driven approach is ultimately what we ended up with. But it did take us some iteration to get there because I think the biggest challenge is coming in. You see what every other peer is doing in this space, and you're like, oh, if Facebook's launching it, it must be working because they have millions of users, and their growth is great. But the big mistake there is you don't know the circumstances of their product.
00:05:44 · Sandy Diao, Guillaume Moubeche
You don't even know if they're measuring the success of these emails. Right? And so I I think it kind of took us several iterative cycles to figure out that more of a data driven transactional approach was gonna be much more beneficial to our growth. [00:05:58] S1: So your focus at the time was how do I connect activation with retention. Is that correct? [00:06:06] S0: Yeah. It was retention at the time. We weren't monetizing quite yet because we were pre revenue. We hadn't built out our ads platform. But I think, you know, it was retention in the in the sort of North Star metric view. In the short term, it was, can we get a user to essentially come back and look at, you know, meaningful day one, week one, month one retention? And then, of course, how does that translate into downstream month six, month 12 retention? [00:06:31] S1: And typically, in the case of Pinterest, I'm making that up, but let's assume that when a user sign up, as you mentioned, they need to, let's say, create their first board, start, like, peeling things, etcetera, etcetera. And once they do that, we believe that they're gonna stay longer. So as as a person working in growth, you have, I'm guessing, like, a list of hypothesis that you wanna test. How exactly do you convince the product team to change things around? So eventually, you know, like, you get to a point where your growth experiments can lead to higher retention?
00:07:09 · Sandy Diao
[00:07:09] S0: So that's a really interesting question because at the time at Pinterest, we didn't really have... At least when I joined because I was so early, we didn't have a product growth team. We didn't really have engineering scope to make some of these changes that would actually drive incremental jumps in, like, conversion rates or activation rates. So I have kind of a a funny story on how I managed to convince an engineer to moonlight this, which is at Pinterest early on in our culture, we, actually had a lot of people who like to play foosball. Are you familiar with foosball, the tabletop soccer name? Yeah. So I noticed that one of the front end engineers who would be able to control some of the onboarding surfaces actually play foosball during lunch. I wasn't necessarily a big foosball fan, but I thought, oh, wow. This is like a good way for me to get in front of this engineer. And so I joined one of the lunch foosball games one day, and I just introduced myself. And then I kind of started to like foosball because it was competitive. It was kind of a nice break from work. And then I ended up staying... This is funny. I haven't really shared this with anyone before, but I stayed after work every day to practice foosball. And I became decent at it, so much so that I ended up playing a lot more of the games. And in one of the games, when I was playing foosball, I mentioned to this engineer that, you know, oh, we have this issue. I don't know who to go to. And he said, oh, you know what? I touched that surface. I have the code there. You know, why don't you draft something up? And even after, you know, that discussion, we had a formal meeting. We found out that it wasn't gonna fit in a proper road map because there were so many other competing priorities and new features to build. But you mentioned, you know, I'll stick behind after work a few days a week, and, you know, we'll moonlight this thing, and we'll test it and see if it works. And so that's actually how I got, you know, the first iteration of the onboarding, emails and email... And onboarding experiences to be coded into the product itself. So I think every company, you know, big or small, has a unique culture like that. But I I think in a lot of growth roles, that's kind of how a lot of big things tend to happen, which is it doesn't get planned for. It's opportunistic.