Match Group, Inc Citi’s 2026 Global TMT Conference

In August, we're getting close, closer by the minute. In August, it was down minus 1%. Better than July and better than Q2, so we're headed in the right direction. September's looked a little bit better so far too. We're getting there, and we should get there soon. Really, the biggest factors are the algos, which we've talked a lot about. It's a culmination, and we've done a lot. It's all the product changes, but if you had to boil it down to the one, it would be the algos. In particular, we released a new unification algorithm, the version 2 of that algorithm, mid-July. That's had a pronounced effect on sparks and sparks coverage, all of which have been better in August. That's leading to the improvement in DAU too.

I do think the DAU and MAU metrics will ultimately converge. I think what you're seeing now is there's really two drivers of MAU. There's users, new users, and then there's retention, right? The product of those two things is effectively MAU. Why you're seeing DAU outpace MAU today is we've made more progress on the product side, on the engagement side. Users are staying longer, and they're using the app more frequently throughout the month because they're getting more value out of the experience. That's what's improving DAU at a little faster clip than MAU.

The big push over the next couple of quarters is around reconsideration and perception change and getting that other part of the formula, the user growth part of the formula, headed in the right direction so that that's how we get MAU from the minus 7 we were at in Q2 up to flat. I think over time, throughout 2027, those two metrics will converge.

There's really two parts of it. One is technical, which is just the approach to the algorithms. Spencer talks about it this way, I think it's a good way to lay it out, is before, the old architecture was, think of it as stacks of cards. There's about 15 stacks of cards, independent stacks of cards being profiles. Each stack was optimized towards one objective. So it might be the revenue in a stack optimized towards generating revenue or generating matches or generating a boost for the new user experience. Then we basically decide which of the 15 stacks we would show you first. You'd get through that one stack, and then we decide what next stack to have you go through. Now it's one unified stack.

There's one stack of cards, not 15, and that technically has just worked better and been more efficient and effective. The second part is really not technical, it's more strategy-driven, right. That is all around optimizing this new unified algo towards sparks and towards meaningful connections. Whereas before it was, again, a number of different constraints, revenue or matches or what have you, and that's been a huge unlock. I think that's probably been the most powerful thing. One of the things, I'm sure we'll get a question on it, is around the user give back budgets. One of the reasons we created that budget, we were a little concerned that we might end up spending $60 million in revenue on user give backs. It ended up being less, is because we removed the revenue stack altogether.

We thought that would have a pretty detrimental impact on the business. It turned out not so much just because this unified stack as a whole just outperformed those individual stacks, even when we deprioritized the revenue.

Yeah. We do have a team. We have a central AI team that's working on more longer-term horizon efforts that are centered around some of the things you just mentioned. Really leveraging the size and the scale of our business in more a unified data stack way that we can apply AI and machine learning to. That's a little bit more longer term. I think the next big unlock for Tinder specifically when it comes to algorithms is more real-time algorithm updates. So later this year, probably towards the end of the year, later in Q4, we're going to roll out real-time matching or real-time algo updates. Today, it takes about four hours for the algorithm to adjust to your changing behavior on the app. That's far behind, I would say, the benchmark today for a social media app.

We've all used TikTok or Instagram, where it feels like it's being updated in real time.

Within a session. Within a session.