Marketing mix modelling: what each channel adds, week by week.
A marketing mix model uses your weekly sales and spend history to estimate what each channel adds. That includes channels people rarely click, like TV and video.
For brands spending across several channels, with at least two years of weekly sales and spend data.
- Free two-week audit
- Built on your own sales data
- Checked against your tests
Three things from each model.
What each channel adds
Sales from each channel each week, with a range, next to the sales you would make with no ads.
Where more money stops working
For each channel, the point where extra spend brings in less and less.
A budget plan
How to split next quarter's budget, and which result a test should check first.
Two weeks to the first answer.
- Audit, weeks 1 to 2
We check your data: enough history, no gaps, and big events like price changes written down.
- First model
Usually four to six weeks. We compare it with any tests you have already run.
- Every quarter after
The model is refreshed with new data, and tests check its biggest claims.
Your history and one owner.
- At least two years of weekly sales or sign-ups
- Weekly spend by channel for the same period
- A list of price changes, launches and other big events
Common questions.
How is this different from attribution?
Attribution follows clicks. A mix model uses weekly totals. So it can also estimate channels with few clicks, like TV and video, and the sales you would make without ads.
How much data do we need?
At least two years of weekly data, so the model can see the seasons, and more if your data is not split by region. With less, we tell you how sure the result can be.
Can we trust a model?
On its own, only so far. We check its biggest claims with tests and update it when the results disagree.
Start with the audit.
Two weeks in your data, free. Yours to keep.
Book your free audit