Judge ads by how many trials pay, not by how cheap the trials are

David BarnardYeah. You mentioned earlier the trust side of things and how important trust is and the research that Jeremy's involved in that lends trust to the YouTube channel and to the app ultimately. But how do you think about the creatives and the way they shape the perspective of people coming into the app as far as over-promising or things that are going to get people to convert, but then maybe quickly churn or whatever?

Ethan EthierThat's a big reason why we decided to actually do Meta in-house rather than choose an agency is because we felt like we needed someone who really understood the product and was very passionate about the product and ultimately could promote the app in a way that isn't faking anything. So we wanted to make sure that whatever we put out there actually relates with what the product actually does. I'm sure you've probably seen a lot of companies that over-promise what we can actually deliver. And for us, we just wanted to make that experience from marketing to the product handoff as seamless as possible.

David BarnardAnd have you seen certain ads perform poorly based on that hypothesis or have you sometimes seen the opposite where you see an ad really take off and then you're kind of like, "It's doing really well, but I'm not sure that's the message we want to send?"

Ethan EthierYeah, definitely. We've tested a lot of ads that are more like, I guess, click baity-style or encourage a click and encourage someone to start a trial, but we can quickly see from the trial to paid that this is not the user that resonates the best with the app. So we'll see a huge drop in trial to paid, whereas maybe we think a lot deeper about who this person we're trying to target is, what their pain points are, and then how the app uniquely solves those pain points, that brings in a much higher quality of traffic. So it's definitely a balance of, "Okay, we want people to start the trial at the cheapest cost, but also we don't want just anyone to start the trial."

David BarnardWhat are you targeting against and what kind of data are you sending back to Facebook or Meta for the targeting?