A 20% reduction often changes marketing-mix structure.
A 20% reduction often changes marketing-mix structure. At this scale, unprofitable final tranches are not enough: cuts can cross coverage floors, break contracts, remove learning or eliminate creative capacity. Compare expected and stress loss, protecting items whose reconstruction is slow or costly. The selected plan need not minimise central loss; it must remain acceptable in plausible scenarios.
Decision and method.
Choose a robust cut whose maximum loss stays below approved tolerance, even if expected loss slightly exceeds a concentrated plan. Split spend into tranches and identify floors, contracts, dependencies, teams and exit costs; mark cuts that eliminate a whole capability. Estimate contribution loss in central, weak-effectiveness and competitive-response scenarios, adding reactivation delays and cost. Compare savings, expected loss, maximum loss and risk concentration; exclude plans that breach tolerance. Start with reversible tranches, observe leading signals, cross structural steps only after review and retain a correction reserve.
Worked example.
Three plans each save €200k from a €1m budget. Concentrated cut loses €88k central and €180k stress; robust cut €104k central and €128k stress; pro-rata €120k and €150k. Central cost of robust versus concentrated is 104 − 88 = €16k. Stress protection is 180 − 128 = €52k. With maximum-loss tolerance €140k, only the robust plan meets the constraint. Retain it: it gives up small central optimum to preserve capabilities and sharply reduce loss when assumptions deteriorate.
Checks and limits.
Identify cuts crossing structural steps; calculate loss under at least one joint stress scenario; include reconstruction cost and delay; keep maximum loss below approved tolerance. Stress scenarios do not cover every shock. Competitor reaction is hard to quantify. Sequencing can conflict with a cash emergency.
Resources and sources.
Download 20% reduction scenarios. Related: build a marginal cut, limit regret, formalise floors. Sources: Ben-Tal, El Ghaoui and Nemirovski (2009); Bertsimas and Sim (2004); Keeney and Raiffa (1976).