Reallocate €1.2m without increasing the budget
Reference case
This fully reproducible fictional case uses 156 weeks of simulated French data across paid search, paid social, video, television and retail media. Noroa is not a real company; no result presented here is observed market performance.
The decision
Keep the quarterly budget at €1.2m and compare the historical allocation with a constrained reallocation. The objective is post-media contribution, not platform-attributed revenue or ROAS. The central scenario moves €270,000 away from channels close to saturation towards video, television and retail media; it adds no budget.
Data and validation
The weekly file contains 15 decision-relevant variables: revenue, post-media contribution, media investment, price, promotions, availability, store coverage, seasonality and calibration events. It reserves 144 weeks for estimation and 12 weeks for temporal validation. The published 80% predictive interval covers 11 of the 12 held-out observations.
Decision model
Each channel uses a saturation curve, Rᵢ(s) = Aᵢ × s^αᵢ ÷ (Hᵢ^αᵢ + s^αᵢ), plus an explicit confidence level. Allocation changes in €10,000 increments within published operating bounds; the comparison is between marginal returns, not historic averages.
Conditional result
The target is applied in two stages: six pilot weeks, two reconciliation weeks, then a week-nine decision. Move to the target only when expected contribution is above €30,000, availability is above 95%, price and promotions remain comparable, and no critical margin deterioration appears. Hold the pilot if the interval includes zero and the prudent loss remains below €20,000. Revert if the prudent bound falls below −€40,000, availability drops below 95%, comparability fails or a delivery constraint is exceeded.
Limits
The data, calibration events and gains are simulated. The response curves do not estimate creative or sequential synergies; agency, production and transformation costs are simplified. The minimums, maximums and confidence levels are local to this Noroa case and must be re-estimated for any real organisation.
Reuse the protocol
Use the published data, model parameters and decision note to inspect the assumptions, reproduce the scenario and adapt the stop rules to your own constraints. The case demonstrates a decision method; it does not demonstrate a market.