MMM

MMM - Marketing Mix Modeling, an econometric method for assessing the effectiveness of marketing channels. Unlike MTA (which tracks the path of a specific user), MMM works on aggregated data: it takes weekly or monthly data on expenses by channel, external factors (seasonality, holidays, competitors) and sales - and builds a regression model that explains what brought in what.

The main advantage of MMM is that it does not depend on cookies, pixels and tracking. In a world of cookieless and iOS ATT, where 40-60% of mobile conversions are not attributed to tools like GA4, MMM remains a method that evaluates the true incremental contribution of channels, including TV, OOH and other offline formats.

Historically, MMM was a tool for large FMCG companies with budgets from $10M+ - building a model is expensive and time-consuming. Now more accessible tools have appeared: Meta Robyn (open source in R), Google Meridian (open source, 2024), Lightweight MMM from Google. The entry threshold has been reduced to companies with a marketing budget of $100,000/month or more.

The limitation of MMM is inertia. The model shows the average contribution of a channel over a historical period (usually 2-3 years of data), but does not respond well to new channels or sudden changes in the mix. Therefore, MMM works well as a strategic budgeting tool rather than a tactical dashboard.