£4M
revenue uplift from budget reallocation
Saturation curves fitted per channel revealing diminishing returns
Scenario planning tool delivered for real-time budget trade-offs
Spend data standardised across all ATL channels
Delivered a £4M revenue uplift for an international stationery retailer by building a Media Mix Model that decomposed sales into baseline demand and channel-level incremental contribution. The model enabled evidence-based budget reallocation across TV, radio, out-of-home, print, and digital display without increasing total spend.
The Problem
An international stationery retailer was spending tens of millions annually across above-the-line channels: TV, radio, out-of-home, print, and digital display. Budget allocation decisions relied on gut feel, historical precedent, and vendor-supplied metrics that were neither independent nor comparable across channels.
The Solution
We led the design and delivery of a Media Mix Model (MMM) to replace heuristic-driven allocation with a statistically grounded view of channel performance. The model decomposed total sales into baseline demand and the incremental contribution of each paid channel, accounting for adstock effects (the carry-over of advertising impact beyond the exposure period) and saturation (diminishing returns at higher spend levels).
Raw spend figures do not capture how advertising actually works. Adstock modelling captured channel-specific decay rates (TV tends to have a longer tail than paid search, for instance), while flexible saturation curves (Hill functions) fitted to each channel revealed where spend was productive and where it was hitting a ceiling. Baseline decomposition explicitly separated seasonal patterns, promotional effects, and organic demand, ensuring channels did not absorb false credit for sales they did not generate. Results were presented as ranges rather than point estimates, letting stakeholders reason about trade-offs rather than fixate on numbers that might shift with the next data refresh.
The response curves fed directly into a scenario planning tool. The marketing team could ask questions such as "if we reduce TV spend by 15% and redistribute to digital display, what is the expected revenue impact?" or "if we hold total budget flat but reallocate based on marginal ROI, how much incremental revenue could we unlock?" Each scenario produced a range of expected outcomes with explicit assumptions, turning budget conversations from opinion-based debates into structured comparisons of quantified trade-offs. Alongside the model, we worked with the client's data team to standardise media spend records, fixing inconsistent and incomplete logging that was a prerequisite for credible modelling.
Results and Impact
| Metric | Value |
|---|---|
| Bottom-line impact | £4,000,000 |
| Channels modelled | TV, radio, out-of-home, print, digital display |
| Baseline separated | Yes: seasonal, promotional, and organic demand isolated |
| Saturation curves | Fitted per channel with diminishing-return thresholds identified |
| Scenario planning | Live tool for budget reallocation trade-offs |
| Data improvement | Spend logging standardised across all ATL channels |
The £4M figure represents the revenue uplift from reallocating budget away from oversaturated channels into those with higher marginal returns, without increasing total spend. The gain came from working smarter with the same budget, not from spending more.
Key Takeaways
-
Response curves make budget debates concrete. Once stakeholders could see where each channel's curve flattened, the conversation shifted from "should we spend more on TV?" to "at what point does the next pound on TV return less than a pound on digital?" That specificity changed the quality of decisions.
-
Timing matters more than complexity. Delivering the model output two weeks before the annual planning cycle had more organisational impact than any additional modelling refinement would have. A good-enough model at the right moment outperforms a perfect model that arrives after budgets are locked.
-
Baseline decomposition is non-negotiable. Without isolating seasonality and promotions, media channels absorb credit for sales they did not cause. The decomposition step is what separates a credible MMM from a misleading one.