The marginal dollar.

Budgets are set by last year’s split and defended by averages. The only number that should move money is what the next dollar returns.

Blended ROAS is the most flattering number in marketing. Average a channel’s early, efficient spend with its saturated tail and the report looks healthy long after the last million stopped working. Averages defend budgets. Margins move them.

Averages defend budgets. Margins move them.

Allocation is a curve, not a split

Every channel, audience, and tactic has a response curve, and the honest question at any budget level is the slope where you stand. We estimate those curves from the strongest evidence available — experiments where spend is large enough to deserve them, models calibrated to those experiments where it is not — and allocate to equalize marginal contribution across the plan. Money leaves the saturated tail of a strong channel and funds the steep early curve of an underfed one, which an average would never permit.

The re-cut runs on a cadence

Annual planning assumes the curves hold for a year. They do not — auctions move, creative fatigues, seasons turn. So the budget is re-cut on evidence at a set rhythm, each line entering with kill criteria attached and each re-cut logged against the ledger. Floors protect commitments; ceilings stop overspend past the crossover where the next dollar returns less than it costs.

The plan, in other words, is a hypothesis. The ledger is the test, and the marginal dollar — wherever it earns most — is the entire strategy.

← All Pareto Insights