The Incrementality-to-Budget Decision Framework

INCREMENTALITY-TO-BUDGET WORKSHEET — original Growthcraft synthesis
Decision / request:
Intervention and counterfactual:
Study reference, analysis owner and approval date:
Measured outcome (net revenue, not attributed sales):
Currency, audience, markets and outcome window:
Refund maturity and excluded outcomes:
Estimate / uncertainty method / sensitivity assumptions:

GATE 1 — EVIDENCE
Was this a causal study or an attribution / instrumentation report?
Assignment or identification method:
Contamination, missingness and concurrent-change review:
Status: pass / unresolved / fail. Evidence link and owner:

GATE 2 — SCOPE
Does the proposed decision match the tested audience, dose and period?
Where is it an extrapolation?
Status: pass / unresolved / fail. Evidence link and owner:

GATE 3 — ECONOMICS
Incremental media cost S:
Other incremental cost O (exclude costs already in margin):
Pre-media contribution margin m:
Low / base / high incremental net revenue assumptions:
Contribution profit = incremental revenue × m − S − O
Break-even revenue = (S + O) / m, if m > 0
Status: robust within assumptions / sensitive / adverse. Finance owner:

GATE 4 — ACTION
Decision: hold for evidence / reject current proposal / bounded follow-up
Proposed budget cap (chosen by owner, not a benchmark):
Outcome and customer-quality guardrails:
Review date and mature cohort definition:
Rollback trigger, execution owner and approver:
Unknowns that could reverse the decision:

Never infer causal proof from this worksheet. Never scale solely from average iROAS.

What this framework does

A positive lift estimate is an input to a decision, not a spending instruction. This original Growthcraft synthesis separates four questions: is the evidence credible, does it match the proposed action, do the economics work, and who will own a bounded next step? It is an operational review worksheet, not a validated scoring model or a substitute for experimental design.

Use it when a performance lead, analyst and finance partner are reviewing a completed incrementality study. Do not use it to certify a before-and-after chart, set statistical significance after seeing the result, or justify a national rollout from a small test without examining transferability.

Bring the right evidence

Bring the study protocol, analysis output, outcome definitions, tested spend change, reconciliation of net sales and refunds, and the proposed budget action. Assign an analysis owner, finance owner, execution owner and decision approver. One person may hold several roles, but each responsibility must be explicit. If a causal estimate is unavailable, label the exercise a scenario and keep the evidence gate unresolved.

Four gates, not an average score

1. Evidence: the analyst documents the counterfactual, assignment or identification method, measurement gaps, contamination and uncertainty. A pass requires a review of the actual study, not a platform label. If identification is unresolved, stop the spending recommendation and request the missing analysis.
2. Scope: the execution owner maps tested markets, audience, spend dose, outcome window and product mix to the requested decision. A difference does not automatically invalidate the result; it creates a named extrapolation. Keep the action within the supported scope or design another test.
3. Economics: finance converts incremental net revenue to contribution before media, then subtracts the tested media cost and non-duplicated incremental costs. Examine low, base and high assumptions. Use zero contribution profit as an arithmetic break-even point, not as a universal investment hurdle; finance may require a buffer or a different opportunity-cost threshold.
4. Action: the approver chooses hold, reject the current proposal, or a bounded follow-up. Specify the spending cap, monitoring owner, guardrails, outcome maturity and review date. A positive average return does not establish the return on the next unit of spend.

Do not average these gates into a percentage score. Strong economics cannot compensate for missing causal evidence; a sound study cannot make an out-of-scope extrapolation automatically safe.

Worked example: a profitable base case with a fragile decision

Synthetic example, not a client result. A completed study concerns a specific four-week media intervention. Its hypothetical incremental revenue scenarios are €18,000, €30,000 and €42,000. Media cost is €12,000, extra creative cost €2,000, and pre-media contribution margin 60%. Contribution after those costs is −€3,200, €4,000 and €11,200. The continuous break-even revenue is €23,333.33 recurring.

The analyst has reviewed the study, but these three revenue amounts are chosen sensitivity inputs, not a confidence interval. Finance marks the economics sensitive because the downside loses money. The proposed request is a doubling of spend in new markets, outside the tested scope. The worksheet therefore does not approve the rollout. A useful next action is a separately approved, capped follow-up in a supported audience with a new measurement plan. Its cap must come from the owner's risk budget; this framework supplies no invented “safe” percentage.

Failure modes and limitations

• Instrumentation masquerading as growth: better recovery of conversion signals can increase reported conversions without proving that the intervention caused additional sales. Keep that diagnostic separate.
• Double-counted cost: media belongs in the explicit cost line here, not also inside the margin. Deduct returns from revenue or margin consistently, not twice.
• Mismatched scale: a study estimate for a tested subset cannot be combined with national spend. Reconcile the numerator and denominator before calculating.
• False certainty: naming three assumptions “low/base/high” supplies neither a probability distribution nor coverage. Preserve the original study's uncertainty separately.
• Selective evidence: preserve the initial question and protocol. Do not quietly drop an adverse market after results arrive.

Method background

The IAB and IAB Europe distinguish causal incrementality from attribution and describe why method choice must follow the business question. The worksheet adds an original operational handoff and finance review; it is not an IAB-endorsed method. IAB / IAB Europe guidelines, November 2025.