AI will touch every stage of marketing mix modeling (MMM), from cleaning data and building models to translating complex findings into plain English. But speed and automation alone won't make measurement more valuable.

The bigger opportunity is using AI to help marketers get from measurement to decisions, and ultimately action, while keeping humans in the loop.

ROI is not the objective

ROI is one of the biggest reasons marketers turn to MMM. Focusing too narrowly on it obscures the rest of the value measurement provides.

“When we talk about the role of marketing mix, everybody thinks about ROI or return on ad spend (ROAS),” said Tsvetan Tsvetkov, senior vice president, head of global measurement consulting at Circana, speaking at the EMARKETER Future of Digital summit last week. “The first thing we tell our clients when we do an MMM is ROI is not the be-all and end-all. It's a great reason to do the analysis, but it's only one of the outputs.”

At its core, MMM connects marketing activity and exposures to a business outcome, which varies by advertiser. The process generally has three stages: assembling and preparing the data, conducting the statistical analysis, and turning the results into insights that can be acted on.

AI is already making inroads across those stages, but it has moved fastest at the end of the process.

“Obviously, being able to take the results and interpret them in a way that's with you know LLMs and other approaches to be able to get them into plain English for your leadership, right?” Tsvetkov said. “Being able to take what comes out of a really complex statistical analysis and translate it into business terms that everybody would understand.”

AI’s next opportunity is upstream

The bigger changes may come as AI moves earlier in the MMM process.

Data preparation eats 30% to 50% of the MMM process, according to Tsvetkov. AI can automate more of the work required to clean, process, and validate that data, not just checking whether it is internally consistent, but determining whether the numbers themselves make sense.

“We have analysis from previous periods, and we know what a typical spend is for a company of a certain size,” said Tsvetkov. “So, if we get a number that's out of range, AI can tell us, ‘Hey, this is not looking right. Go and fix it.’”

AI can also reach into the modeling itself, folding normative information and historical expectations in alongside the underlying data. That matters because effective MMM isn't simply about producing a statistically valid answer. The results have to make business sense and be usable by the organization.

But greater reliance on AI creates another challenge: trust.

“To me, the challenges are very much related to the level of transparency that companies are looking for in the analysis process," Tsvetkov said.

Humans still have to sell the answer

The human role matters most once insights leave the model and enter the organization.

That means marketers and analysts will increasingly serve as agents of influence: taking AI-assisted findings and determining who needs to hear them, how they should be communicated, and how to push recommendations through the organization into action.

“A lot of what humans would bring to the table is this ability to influence the conversation and ultimately drive the activation,” said Tsvetkov.

That distinction also changes what marketers should expect from their MMM partners. Rather than another dashboard, they need tools that let them ask questions and get usable answers.

Tsvetkov identified three things marketers should look for in an MMM partner: coverage, transparency, and activation.

Coverage means having sufficient high-quality inputs. Transparency means understanding how those inputs become results, particularly as different modeling approaches can produce different answers. Activation means turning those recommendations into changes in marketing plans.

The real promise of AI

Speed is one of AI’s biggest selling points in measurement. But the goal isn’t getting an MMM analysis back sooner.

“Speed is the current focus of the marketplace, and it's not just speed to getting the results; it's speed to making decisions,” said Tsvetkov.

Getting there will require marketers to think of MMM less as a periodic analysis and more as an ongoing organizational program. Data, agencies, measurement partners, and internal stakeholders all need to feed into a closed loop in which recommendations are implemented and their effects measured.

That feedback loop will also have to account for KPIs beyond immediate ROI, including customer acquisition and longer-term effects.

The rise of agentic commerce makes that broader view more urgent still.

“For MMM to really do its job well, it needs to be kind of implemented at the organizational level,” said our analyst Arielle Feger. “You have to have your data there, and in the pipes. You have to have your partners who are transparent and who are actionable, and you have to continually optimize and make sure that it's continually working.”

Watch the full session

This was originally featured in the EMARKETER Daily newsletter. For more marketing insights, statistics, and trends, subscribe here.