Check 2,000 AI prompts, guess their volume from Google, and find the cited pages

JosephIt it's basically the same way as like regular UGC works, like you want to pay high CPM in the beginning, and then you switch to base as you like find your winners. Do you have like AI search stuff, too?

Matt Espinoza… for consumer apps. And to be able to rank on search isn't really that hard hard of a science. There's really two approaches we end up kind of uh playing into the system, as one of the Clover products, the one we kind of started off with was AI search, and then since then we've moved over to TikTok and Instagram as new channels. But we started off with just like uh mass publishing on Reddit and blogs and been able to get tons of data. Uh And so right now, just on AI search, we generate about 150 million views across all customer bases, and that's a mix of blogs or Reddit content that we do for customers. But to be able to do the process well is really around like three to four steps. The first thing is that I would search uh 2,000 prompts, um and then I would determine, one, what the search volume of each of those prompts are, uh and then, two, uh the citations of those prompts. So you can imagine, let's say you're an app that's doing food uh tracking, uh you'd probably want to rank for what is the best app for food tracking, and then you can determine that that particular search term on Google might be doing 15,000 impressions per month for that particular search query. You can then assume, what we typically assume, is a 30% rate for AI search. And so in this case, it would be around, let's say, 5,000 uh impressions particularly just to that same search query on AI search. And so, this is like still a very viable set, cuz you can imagine that's just one prompt for one search query, and if you do that 2,000 times, you can find some really good uh nuggets that are really worthwhile to to do. But generally speaking, the Google, you know, representation of search can just be almost attributed to AI search, cuz there's no good prompt volume that exists online, so you kind of have to do estimates on what exists on existing search queries.

JosephAnd there's like a lot of existing SEO tools for that uh where you can just like you search up any SEO tool and then you just find like regular Google search queries uh you find the volume on that, and you estimate 30% of that volume is probably going to LLM searches, then. Interesting. Is there like a particular method you like to use to find those queries?