The Exchanges

Q A sign up is a new lead, right? You then sell to a diabetic company or any other company.

A So, so let me get going on that. Basically what we ended up doing is we ended up taking that question asking model and just scaled it way up. So now we have hundreds of questions. We obviously don't ask everybody a hundred questions. It's usually somewhere between eight and 10 questions, but literally we took that model and we literally just ask people questions up value the lead first, right? Because what the questions really allow us to do is to find data points. Like, are you diabetic? And if they are a diabetic, suddenly the value is just massively greater. And so we do that For car insurance, we do that in dozens of different spaces. And so, yeah, that's exactly what we still do. And so, so the lead value is really more of a composition of all your answers than actually just like one lead. But yeah, that's still pretty much the model.

AI assessment note: “And so, yeah, that's exactly what we still do.”

Q exclusive content, or take the quiz to see if you're smarter than Donald Trump. Right, when they enter that quiz, it's gonna say, uh, I'm making this up, Donald Trump is, Trump is diabetic, are you? Right, like, you, you frame all the questions in the relativity of, like, whatever the, the, the headline was. Why go, why are you going the route of just saying, click here to register?

A Well, so a, it's going to be much more complicated to custom write the questions for each of the articles. So the questions are built to be more broad. Um, would it work in that context? Absolutely. So I think it has to be enough of an incentive though for users to actually sign up. So we basically see one single content piece is potentially not enough of an incentive to sign up. I think, right. And I think to the company itself, Um, people probably don't want to take a survey 12 times, right? They don't want to come back, and every time they see an article, they want to take a survey, right? That's going to just end up being the ad model all over again.

AI assessment note: “it's going to be much more complicated to custom write the questions for each”

Q Click here to register, and then, okay, so they click register, and then would you frame the questions around entrepreneur-related content, or would you put up your generic question set?

A So both, actually. So we can do, we can do custom question sets that really fit into the website, and so what we have done, we've built a ton of models, actually, that figure out which questions work best on which property. So the goal is that the questions auto-optimize themselves, and that they just Basically optimize themselves for the best response rates. So what we have seen is that when we put it onto a new property, some questions that work really well on old properties don't work as well on new properties because either the audience is different or the perception of the questions in the context is slightly different. So, but the beauty is we have built a system to where it auto optimizes itself where we don't write. The goal is not to you as a publisher to have to write 15 different questions, right? The goal is that you literally just put in a piece of code and I send you a huge check every month. Right.

AI assessment note: “So both, actually. So we can do, we can do custom question sets”

Q Okay, so a little more expensive, I'm just kidding. All right, good. Okay, so tell us, what is Zito? What does it do, and what's the revenue model? How to make money?

A Right, so we basically took the question model that we built under samples.com, so we, we spent the last five years building tech around that, but so, and really, I mean, so it turns out, it ends up being really, really hard to ask those questions in the right order, and as efficiently as possible, right? If you have three questions, yeah, that's fairly simple, but if you have hundreds of questions, figuring out which one to ask when, and then which advertiser wants to actually buy against that, right? So it's literally just an ad inventory play that is triggered by questions, and so, So what we did on the Zito side is we built a ton of tech to actually make that much more efficient, and so now the play is really that we're taking the same model that samples.com, what would make samples.com so successful, right? I mean, the tech had a large piece to do with that, especially in the beginning. And we're just letting other publishers access it, right? So many people have a hard time monetizing their traffic, and so we're literally just building, um, it's kind of similar to Google AdSense. It's just an ad tag, basically, that we give you, uh, but the ad tag asks questions first, up values your traffic, and then sells ad inventory against it. To give you an idea, right, regular CPMs from AdSense range from five to 20 dollars. And by the way, AdSense is a great product. We, we love i…

AI assessment note: “an ad tag basically that we give you uh but the ad tag asks questions”

Q Why doesn't every company see what you saw and create their own little free sample, free giveaway, collect their leads, ask good questions and, and build their whole onboarding funnel around that. Why don't people do that? You think it seems like such a much smarter way to do business.

A Yeah. Well, I've asked that myself quite a bit. Um, so a, it does actually turn out to be really quite hard technology speaking. Um, So the math is quite complex, right? I mean, we were talking about seven billion possible combinations as kind of like the current, the current stage, right? Then you need to find all the advertisers that want to buy against that. And so that's kind of where we see our opportunity is that we can offer a product that gives people really the ability to use questions without having to hard code them, right? When you hard code that stuff, it's just the efficiency is fairly low. And I mean, we spent the last five years getting away from having to do that and really getting those efficiencies They'll figure it out.

AI assessment note: “a, it does actually turn out to be really quite hard technology speaking”

Q I'm going to cut 80% of the workforce. I'm going to keep the best editors that create the best content, not clickbait, and I'm going to say Stefan is the guy in charge of monetizing all of this traffic. You're going to basically put together some kind of article that gets people into a survey. What's the headline on that article that we run on the front page of newyorktimes.com?

A Yeah, so, so I don't know if I would cut all the content, because they have some pretty amazing content. So I think how we're thinking about that specific example, right, is the New York Times, basically, you end up having a paywall, right? So that's part of their revenue model is ads, part of their revenue model is a paywall. And so we really see ourselves more as a, maybe not a paywall replacement, but as a much better addition to that, right? For example, you say, hey, you've seen your 10 articles, you have to pay, right? Um, our offer is basically, hey, either pay or register, right? The registration will be, have some inherent value to New York times to begin with because they can remarket to it, right? But we can then also monetize them and pay them two bucks per registration, right? It's a much lower friction way of getting some extra revenue from each user and getting it much quicker. So it's really more, so it'd be more of a, we're not the only revenue source for a company, right? We're incremental and we really add in pretty well with the existing ad model. And even with a paywall model, it's just kind of a nice alternative to have to test.

AI assessment note: “So I think how we're thinking about that specific example, right, is the New York Times”