Ars X Machina’s agile mix modeling penetrates the walled gardens to measure them against other channels
By Michael Bürgi • October 15, 2025 •
Ivy Liu
This article is part of Digiday’s coverage of its Digiday Media Buying Summit. More from the series →
Media mix modeling (MMM) had its day in the sun when traditional media ruled the roost — then came the relative instantaneity of digital marketing and media, which turned MMM’s lack of timeliness into a significant handicap.
Media agencies have been working ever since to harness the power of machine learning and AI to speed up the process from months to days.
Independent media agency Ars X Machina is taking that effort seriously with the commercial launch of its agile mix modeling proprietary platform, which essentially aims to deliver full campaign measurement across offline and online channels — the full funnel. What’s notable here is it includes the major walled gardens.
While agile mix modeling (AMM) has been in some form of testing for close to a year, it’s formally being rolled out today, and clients including Sierra Nevada Brewing Co. and GE Lighting have made use of it to optimize and adapt their campaigns while still in market.
AMM lets marketers mix-and-match different online and offline media, ostensibly comparing Meta against podcasts, Amazon Ads against out-of-home, and CTV against search — all on a level playing field, allowing budget shifts while campaigns are still in-flight. It can be used to measure everything on a campaign from incremental revenue and ROI, or incremental customer acquisition and cost per acquisition, said Sara Owens, svp of analytics and data science at AXM. “It can handle whatever business metric you want, where you have a time series data set for it,” she said.
“If you can’t measure everything in your campaign — if you’re missing pieces — then your measurement is just going to be wrong,” said Owens. “So we thought, market mix modeling is so smart, but why does it have to take so long with all these advances in machine learning and automated data collection. Let’s see if we can do something faster. So that’s what we built.”
Owens explained AMM employs standup dashboard environments for all of the clients using AMM, that are pulling in delivery data from campaigns through API’s right from all the major ad serving platforms. Then machine learning is used to crunch the data to seek out results.
For Sierra Nevada, some of the early testing has improved some of the brewer’s local market results, delivering ROI increases of up to 40% in some cases.
“There are certain intangibles that you cannot know [that] are informing attribution,” said Will Mestayer, director of connected commerce at Sierra Nevada. “We have so many different channels that we’re trying to pull together, and this has really helped us isolate what’s working in a very specific market where we we had no telltale signs.”
Owens and Mestayer will be speaking about AMM at today’s kickoff of Digiday’s semi-annual Media Buying Summit.
“When we think about working dollars that we’re infusing into the brand, we want those to work the hardest they can for us,” said Mestayer. “So what it allows us to do is really optimize working dollars that are in flight, to drive commercial impact.”
Ars X Machina rebranded just over a year ago from Media Matters Worldwide, a female founded media agency, in part to avoid being confused with Media Matters, an organization that’s been embroiled in legal action with X and Elon Musk.
More in Media Buying
