MMM

Marketing mix modeling (MMM) is a statistical method that uses aggregated time-series data, such as weekly spend by channel, prices and seasons, to estimate each channel's contribution to sales and the effect of budget changes.

MMM is the oldest of the main measurement methods and has returned to favour as user-level tracking has weakened. Meta's open-source Robyn guide calls it an econometric model of the incremental effect of marketing and other factors on a chosen KPI (Meta, Robyn analyst's guide, accessed 9 November 2026). It runs on aggregated data, not user-level data, and covers offline channels such as TV and print alongside digital ones.

Google's open-source Meridian answers three questions: the historical return and contribution of each channel, the response curve of each channel, and how to allocate future budget (Google Meridian, accessed 9 November 2026). Models need a long, consistent run of data, and they struggle when channels always rise and fall together. Calibrating them with experiments, such as geo holdouts, narrows the uncertainty.

Example: A model shows that CTV returns $2.10 in sales per dollar up to $200,000 a week and only $0.60 per extra dollar beyond that. The brand moves its spend above that level into search and retail media, then checks the change with a geo holdout.

Related terms: MTA, geo holdout, incrementality, reach and frequency, iROAS, attribution window.

Authoritative source: Google, "About the Meridian project" (accessed 9 November 2026).