Marketing mix modeling for data-driven investment

Our client, a leading multinational FMCG company, wanted a clear read on how well their marketing spend worked across many brands and markets. The goal was to measure what each channel actually contributed, forecast more accurately, and put future budgets on firmer ground.

What we built

We built a marketing mix modeling framework that spans multiple countries and brands, trained on historical weekly sales alongside media investment, pricing, promotional, and seasonal data.

The models account for how advertising keeps working after it runs, an effect called adstock carry-over, and for diminishing returns, so the estimated impact of each channel reflects reality rather than raw spend. That let us attribute sales contribution channel by channel, across television, digital, promotions, pricing, and other commercial activity.

On top of the model, we added scenario planning and budget optimization, so marketing teams can simulate an investment strategy and see the expected impact before a campaign runs. Model outputs feed into stakeholder dashboards where business users explore the numbers themselves.

The outcome

Marketing ROI forecasting accuracy improved by roughly 15%, which gave the team more confidence heading into each planning cycle. With clear channel attribution across brands and markets, budget decisions shifted from instinct to evidence, and scenario planning helped the team direct spend toward the channels returning the most.