Give each paid feature its own paywall and check its sales every morning

David Barnard… years where you were like hardcore optimizing the revenue of pulling features in and out? So would you put something in the free tier and then, oh, wait, we really should have made that paid? How did you run those sorts of experiments? Because those are hard. And once somebody's gotten something for free, then putting it behind the paywall is weird. And then, oh, I paid for that. Why is that now free? How did you run those kind of experiments?

Jeff MorrisYeah, I think we might've been a bit different from normal companies, but the revenue roadmap and our products were very much built in a different part of the org. We were on the product team, but when we developed a revenue feature, it was kind of thought of as being a part of the revenue product line that we... We actually didn't pull things in and out very often. There were some cases where we thought we could charge for something and then we would look at the conversion data on that paywall and it just wasn't moving the needle and so we would make it free. But we never took something that was free and made it a paid feature as far as I remember.

Jeff MorrisAnd I think in most cases, people overthink the user impact of a lot of these decisions where if you make that decision to go free to pay, you can always reverse that decision. Or maybe you have these moments of where the users are like, hey, I hate that you made this decision. But if you're solving a big enough problem, they're going to stick with your product and so you just need to do so in a way that feels transparent and fair. We always …

David Barnard… testing of paid and then switching some of those to free. What would that look like? It would be like the paywall would very specifically highlight this new feature you built, and then if the conversion on those paywalls wasn't high, then you were assuming that that particular feature wasn't valuable enough to drive enough incremental revenue that then maybe it should just be pushed back into free. So that's how you were testing these things?

Jeff MorrisYeah. Pretty much every subscription feature had a paywall entry point that we had a dashboard we'd wake up to every day and we could see the number of, obviously the number of clicks it got and then look at the conversion rate and then map that to incremental revenue for the subscription product. And so you could see really quickly which products were getting the most attention. And that could have been because the paywall was positioned... We always thought of it as being like real estate. And so you have a small canvas and a small amount of real estate. So maybe the problem is you're not showing the paywall in an aggressive enough way, or it could simply be that people are tapping on the feature and deciding that it's not valuable enough for them to pay for.

Jeff MorrisI think what we quickly learned is it's like any portfolio of companies or products. There's a power law to subscription features where one or two features are probably going to drive the majority of your new subscriptions. And everything else is nice to have, but it's incremental. That could change. I'm seeing a lot more creativity with multi-product revenue lines. Robinhood is a great example. It's not all subscription revenue, but I think they have 10 products that are independently generating north of $100 million each. And so I look at that. But at least when we were thinking about Tinder, it was a lot simpler. We had two to three power features that people really paid for, and then we would try and design more features on top of those over time or new experiments, but it was pretty consistent.

David BarnardYeah. One particular experiment that you mentioned was a single experiment that generated $50 million in revenue. Tell me about that one. I mean, at Tinder's scale, maybe that wasn't... And maybe you can fill in the details there of how big the scale was, but tell me about that one experiment that generated $50 million.