How data, micro-experiments and modeling infuse everything Known does

Known is much more than a data and insights agency. It’s a media, creative and experiential marketing shop that also produces original programming for Netflix, Amazon Prime and Hulu. But data and insights are behind all of that in some way. And when faced with the loss of a big client (Farfetch) at the start of 2024 that represented more than a quarter of its revenue, the agency intensified development of AI and analytics tools, spawning both new systems and a 6% revenue gain. See all of Ad Age’s 2025 Agency A-List winners here Skeptic AI is a tool that’s central to a wide array of Known work. Over the past three years, it’s been trained on data from 1.8 million micro-campaigns, churning out more than 4 billion person-level predictions across more than 100,000 campaign assets. “We’re the dumbest we’ll ever be on day one” of a campaign, said Kasha Cacy, chief media officer of Known, because Skeptic is constantly evaluating performance experimentally—not only of media plans but also creative elements. “We have a proprietary way of tagging creative assets that allows us not to just identify a version of creative but also the headline, the image, the body copy for videos. We can say to a client what asset performed better. We know why, because these three things made it better,” Cacy said. “So when we go into social platforms, we don’t just set the target audience and think it’s going to work. We set up lots of target audiences around that main target.” For media clients, those micro-experiments keep rolling nonstop across all media channels, said Ross Martin, president and co-founder of Known. “But even if we’re not your media agency of record, many clients use this tool alongside their agency of record with their in-house media teams.” Segment Simulator, a Known tool that launched last year, is also widely available across clients, including for hands-on use, allowing them to interact with AI-generated customer proxies. That can include asking a simulation of an individual within a segment how she might react to a given approach, then tweaking things based on the feedback. About a quarter of clients use Segment Simulator directly. And internally, “creative teams love it,” Cacy said. “They all want to use it.” Sometimes, a tool Known develops for an individual client can be applied much more widely. A case in point is the Federal Emergency Management Agency data integration Known developed as media AOR for Moderna after winning the account last year. The data is used to turn off ads in disaster areas, something that’s of wider relevance for most other marketers. Read more: How Known grew after a major client’s collapse Known also last year rolled out Causal Modeling, which identifies upper-funnel brand-building messages that cause lower-funnel actions, such as sales or sign-ups. It provides early notification of when brand messages are likely to lead to sales or other conversion metrics, said Known Chairman-CEO Kern Schireson. But the tool goes beyond just shaping messaging to cover entire media plans and go-to-market strategies. Causal Modeling helped Known find “multiple times over the past 12 months” that the upper-funnel solution might not be an ad campaign at all, but something entirely different, Martin said. “It might be we should develop some original programming for you, a movie or a TV series,” he said. “Or we should change the way your logo and design system looks, because it’s creating the wrong feeling at the top of the funnel and not helping us at the bottom.” See all of Ad Age’s 2025 Agency A-List winners here

How data, micro-experiments and modeling infuse everything Known does

Featured Stories