Marketing mix modeling (MMM)

Marketing mix modeling (MMM) is a statistical method used to estimate the impact of marketing tactics on product sales. It typically applies regression models to sales and marketing time-series data, and is often used to decide how to split advertising and promotional spend to get the best return in sales, revenue or profit.

How it works. A model splits total sales into two parts: base sales — the demand a product would have without advertising, driven by factors such as pricing, distribution, long-term trends and seasonality — and incremental sales driven by marketing. It accounts for adstock (advertising's lingering effect), diminishing returns and the carry-over of past campaigns. The results can be used to run "what-if" scenarios that reallocate a budget and show the likely effect on sales.

Aggregate, not user-level. Multi-touch attribution measures marketing at a granular level; MMM works in aggregate. Google says its Meridian model uses aggregate data and does not need individual identifiers or cookies. Meta says its Robyn package requires no personally identifiable information or log-level data and does not depend on cookies or pixel data.

Where it came from. The techniques were first applied to consumer packaged goods, whose makers had accurate sales and marketing data. Data companies Nielsen and IRI later began bundling an MMM into their standard data contracts.

Open-source models. At the start of 2025 Google launched Meridian, an open-source MMM that uses Bayesian methods and needs at least two years of data. Robyn is an open-sourced MMM package from Meta Marketing Science; it uses ridge regression, among other techniques.