Cutting the budget by 10%: removing the least useful last euros.
Evidence status
What was reproduced, and what it does not prove.
The CSV and the Python run verify the marginal losses and the total cut of €100k in the illustrative scenario. They prove no market return.
- Technical specification of the 10% budget cut, specification in French (opens in a new tab)
v1, 31 August 2026 · CC BY 4.0 · specification in French · Method document written by marketing-mix.net. The figures are illustrative: they are not market measurements.
Key distinctions
Three conditions before calculating.
Avoidable tranches
A removed spend becomes a saving only if the cost can genuinely be avoided within the horizon. Contracts, teams, production fees and commitments must be separated from media or variable activity.
Marginal loss
The channel with the best average return can contain a saturated tranche of little use; the average channel can lose an essential capacity below a floor. The decision ranks the last tranches, not the averages.
Reversibility and learning
An immediately reversible cut does not carry the same risk as a market exit or the removal of a testing programme. The cost of lost information and the time to rebuild must be visible.
Method
Building a budget reduction without destroying the good mechanisms
- 01Break down the spend
Divide each lever into avoidable tranches with date, commitment, exit cost and operational floor; exclude the savings that do not materialise within the period.
- 02Estimate the loss
Apply the response curve, the margin and the interactions to estimate the contribution lost per tranche; present a range when the tranche lies outside the historical variation.
- 03Optimise under constraints
Rank the tranches by marginal loss, then apply floors, synergies, minimum coverage and the ability to reactivate; compare with an explicit pro rata scenario.
- 04Prepare the control
Define the leading signals, the stopping thresholds and the review cadence; keep a small reserve to correct quickly a loss above the expected interval.
The ladder of tranches to remove
Each budget tranche is ranked by expected loss of contribution, with its contractual limits and its level of uncertainty.
The marginal cut preserves €16,500 compared with pro rata
A portfolio of €1M must save €100,000. The last tranches available lose €0.45 of contribution per euro on A, €0.90 on B, €0.30 on C and €0.60 on D. D can only be reduced by €30,000.
Marginal plan: remove €50,000 from A, €40,000 from C and €10,000 from D; loss 22,500 + 12,000 + 6,000 = €40,500. Pro rata: €30,000 from A, €25,000 from B, €20,000 from C and €25,000 from D; loss €57,000.
The marginal cut is retained with a sensitivity range. It avoids €16,500 of expected loss and protects B, whose last tranche remains the most productive.
Acceptance conditions
What must be true to act.
- 01
Every tranche is genuinely avoidable within the reduction horizon.
- 02
The loss is computed on the marginal return and not on the average return.
- 03
Floors, synergies and rebuilding costs are explicit.
- 04
The plan carries monitoring signals and a reactivation rule.
Limits
What this analysis does not prove.
- Historical curves extrapolate large simultaneous cuts poorly.
- Competitors can react to a fall in presence.
- Brand and learning effects are more uncertain in the short term.
References
Works cited.
- Little, J. D. C. (1970), “Models and Managers: The Concept of a Decision Calculus”, Management Science, 16(8), B466–B485.
- Doyle, P. & Saunders, J. (1990), “Multiproduct Advertising Budgeting”, Marketing Science, 9(2), 97–113.
- Keeney, R. L. & Raiffa, H. (1976), Decisions with Multiple Objectives.