Sarah Friar: Great. Thank you, Chole. Appreciate the questions. Let me start taking the first one on overall the ad tech stack. So, front-end and back-end, you talk to Nextdoor Ads Manager. That’s the actual interface that’s where people create an ad and really our work there is all about reducing advertiser efforts. So, today, where we’ve gotten to an advertiser of any size can now self-serve on NAM. In addition, we continue to build out other features and functionality with larger advertisers and their agency partners. So that includes releasing things like user roles and permissions, creative duplication, pixel tooling, audience exclusions, ad credit tooling, multi-edit functionality in NAM. So, you can kind of see the list is long, it gets very, very specific to ads.
But we are making progress every single period to make better and better for businesses of all sizes. On the back-end, Nextdoor Ad Server, as we’ve mentioned, we’re now delivering 100% of all U.S. SMB demand, and we’re seeing faster loading times, better distribution of ad impressions. So, overall, that’s driving better retention of those SMB customers, that’s also driving better revenue, up 23% year-over-year in the quarter. And then, of course, those are places where we do also want to use more ML or AI to make sure that the right ad content is getting in front of the right user, which of course then will drive engagement. Beyond our ad platform, it is also worth noting that we do continue to build out first and third-party app integrations.
So, for example, in the first-party side for measurement, we now see that the majority of our CPA-focused customers have adopted our own pixel. Scripts is a great example of a customer who has used our pixel to better understand neighbor activity to optimize for lower CPAs, and they’ve increased spending really meaningful year-over-year. On the third-party side, we now have measurement with folks like Neustar, Oracle, Lucid, Foursquare, Thesus [ph]. And on the brand safety side, which always remains important in social, we are ready live with Moat around things like Verified impression and we’re in the process of integrating IAS. But of course, 1 of the things we always felt that advertisers, if you’re starting from a point of Verified neighbors.
So, in terms of verifying the impression, we’ve already told you that it’s a real person that we know where they live locally. The second question was around Generative AI. So, from a – probably AI overall, so, first and foremost, AI is already really integral part of the product experience here at Nextdoor, whether it’s about improving core experiences such as you see the notifications and you can see more personnel or around applications that use Generative AI. We’ve talked a little bit about, for example, use of our Post Assistant. So when you’re creating a post on Nextdoor, in particular if you’re trying to look for a recommendation, just this week, I was looking for family photography, or if you want to post about a business saying what a great business they were, our Post Assistant jumps into action for you and creates a post that we know because it’s trained against Nextdoor data will perform better on the platform.
And we see about a 70% increase in – 70% acceptance rate of that post suggestion. We’re also going a lot deeper on our active machine learning models in the product overall. That’s increased – since 2022, that’s increased about 10x. And in fact, the number of successful model experiments has increased by approximately 100 times. So, we are prioritizing the velocity and the breadth of different large language models across markets and across user needs. As I look forward, probably the area I’m most excited about in terms of AI is the ability to leverage that not just within the ad server itself to help advertisers understand audiences better, optimize our ad targeting and so on, but also to think about for those small businesses, how do we leverage Gen AI to help them assist with, for example, their page creation or with a campaign.
