Improve return on marketing spend with Marketing Mix Modelling.

If we cannot find a 20% ROI improvement in 12 weeks, we refund you in full. It applies to the first build on every plan. We offer this because we are confident in the work. If your marketing is already perfect then you do not need our help.

For brands spending more than £1m a year on marketing but reliant on attribution to measure impact. If your platform-reported ROAS keeps telling you everything is working, but your blended return tells a different story, that gap is what MMM resolves.

Attribution credits conversions to the touchpoints it can see, usually within a short window. MMM models incremental revenue across every channel you spend on, including channels with no clickable signal: TV, out of home, sponsorship, organic word of mouth, and brand effects.

The category has matured fast in the last few years. Open-source frameworks like Meta's Robyn and Google's Meridian have made rigorous MMM accessible to teams that previously could not afford a measurement consultancy. Industry bodies including the Marketing Accountability Standards Board and Nielsen have published extensively on why MMM is the right answer when attribution alone is no longer enough.

The two methods answer different questions and the best teams use both. For a longer treatment see Beyond attribution, or compare all three in MMM vs attribution vs incrementality.

Incrementality testing runs a controlled experiment: it holds a comparable audience or region out of exposure and measures the gap, which proves whether a specific spend caused extra sales. MMM works the other way round, using two years of your own weekly data to explain what every channel contributed, including the ones with no click. Incrementality answers one question cleanly. MMM answers the whole allocation question at once, and the experiment can calibrate the model. For the full comparison, see MMM vs attribution vs incrementality.

Two years of weekly spend by channel, two years of weekly revenue, plus any media plan flags (brand campaigns, promotions, price changes, distribution shifts). We handle the data ingestion ourselves through Reni, our proprietary data pipeline. You will never be asked to wrangle CSVs.

The full programme runs in 12 weeks: data ingestion, model build, diagnosis, scenario planning. The 20% improvement target is measured against the model's recommendations within that window. Most teams see initial reallocation insights inside the first 6 weeks.

It depends on the shape of the work and the state of your data, so there’s no price list on this page. The cost guide gives honest ranges and explains what affects the price.

You own the model, the data, and the recommendations. Clients can move onto the Partnership Plan for ongoing refreshes (the model needs to learn from each new quarter of data to stay accurate). You can also stay on Flexi and refresh on demand, or take everything in-house. There is no lock-in.

We are marketers first, so the work starts from the commercial decisions you need to make. Proprietary tech speeds up the work and keeps the price down. We take on two builds at a time, and you work directly with the founder, not a junior account team.

A well-built model predicts your sales from its own inputs to within roughly 5% to 10%. If it cannot, that is itself a signal: something material is driving sales that is not yet in the model, and we find it. Accuracy is the most common worry we hear, so we are blunt about it. MMM is not precise to the decimal, and it does not need to be. It needs to be directionally right about where the next pound should go, and right often enough to grow the business. We pressure-test every model against held-out periods before a recommendation reaches you.

As a rule of thumb, MMM earns its keep once you are spending around £1m a year or more on marketing across more than one channel, with some variation in that spend over time. Below that, the modelling overhead is hard to justify and simpler measurement will do. If you spend the same amount on the same channels every week, the model has nothing to learn from, whatever the budget. If you are not sure where you sit, the MMM Impact Calculator gives you a benchmarked answer in about a minute.