Andrew Boone: I also wanted to ask about MFAs. There’s just new concern in the industry that feels like it came up this last quarter in terms of made-for-advertising sites. Is there any impact that you guys are seeing? Or how does that relate back to Outbrain? Thanks so much.
David Kostman: I think that there’s been a lot of coverage around this topic. I think the topic has been further clarified. First, just on the general statement, it’s very difficult to classify what is an MFA. Many publishers, sort of, their objective is to drive advertising. I think it’s difficult to put all of them in one bucket, the different types of MFA. MFAs, I think many of them, generate real value for advertisers, having real people go to content and advertisers reach audiences that they want to reach at an effective cost. So that’s generally. For us, specifically, we said previously, [5%] (ph) of our ex-TAC revenues. So, it’s not that significant. In terms of driving traffic to MFAs, I think you probably saw the [indiscernible] report, 80% to 90% of the traffic that’s driven to them is generated by social and search.
So again, we — as long as they meet our strict content guidelines and all the other requests that come from security and fraud, et cetera, I think there’s a great category of advertisers that we have. Generally, our focus as a company has been also on helping and supporting quality journalism, premium journalism around the world so we provide vital revenue to them. So, it’s again part of our overall business and combined with our premium publisher business too.
Andrew Boone: Thank you.
Operator: Thank you. And the next question will be from Dan Day, B. Riley FBR. Please go ahead.
Dan Day: Hey, good morning, guys. Thanks for taking the questions. So, I appreciate the update on the algorithm AI bidding strategies there. Just maybe if you have any data points or anything you could share just on the subset of advertisers that you think are using these kind of AI tools better than others? And whether there’s been an uplift in spending performance for them? And then, what you need to do to get those that aren’t using them to kind of start using these AI tools, and maybe that helps increase spend per advertiser for those? So, any thoughts there would be great. Thanks.
Yaron Galai: Thanks, Daniel. Yaron here. So AI, first, as I mentioned in the comments, splits for us into two big buckets. One is the algorithmic AI, which we’ve been building in-house for almost a decade now. And the other is the generative AI. The algorithmic AI are technologies that we deploy across the board. So, they apply to all publishers, all advertisers within our system. We don’t update on the results there all the time, but we did mention last quarter that for the first half, I think we’ve seen an increase through those algorithmic AI changes of potential click-through rate of 4%. And I did mention in this call that we’ve seen in the past couple of months that they’ve been among the three-year record highs in terms of the CTRs. And that again is attributed a lot to all these AI changes, or technologies that we’ve deployed.
On the couple that I mentioned on the call, those are in lab R&D mode. As I said, we are live with them on a few advertisers. What we’re doing with generative AI is really trying to attack ad variety from all directions. So, I mentioned this in a previous call, an advertiser might upload a campaign or even one ad. And through generative AI, through our ChatGPT integrations, we’ll expand that and offer them 200 other variations, which they can decide whether to accept and add to the campaign automatically or not. So, those tools were embedding into the product itself. They are available to ultimately to all advertisers, but it’s really up to them to choose which ad varieties they want to implement and choose. Many of them have been doing it, especially as it relates to creative and headlines.
