Marketing measurement: build evidence proportionate to the decision.
Answer:
Measuring marketing is not multiplying indicators. An evidence architecture links four levels: reliable instrumentation, description of the phenomenon, identification of the counterfactual and transport to a decision. A dashboard can show what happened; it does not establish why. An experiment can identify a local effect; it does not guarantee extension to another intensity. Quality comes from consistency among question, design, uncertainty and financial exposure.
Method and decision:
Classify the decision by money exposed, reversibility, speed, loss asymmetry and waiting cost. Define economic metric, unit, population, window and no-action scenario, then map biases that may separate observation from counterfactual. Choose instrumentation, descriptive analysis, quasi-experiment, experiment or calibrated model; calculate expected power, uncertainty, cost and transport domain. Publish observation, causal estimate, extrapolation and hypothesis as distinct claims, each linked to a launch, stop, pilot, collect or maintain rule. Require stronger evidence as exposure and irreversibility increase.
Worked example:
A €1.2m national campaign is due for renewal. Attribution suggests ROAS 2.8, while comparable-area analysis gives an effect between 0.6 and 1.9. Without further evidence, the probability of funding an unprofitable campaign is set at 35%, with €900k loss. Expected loss without a test is 35% × €900k = €315k. A geo-experiment costs €45k, reduces the probability of a wrong decision to 8% and leaves €72k residual loss. Gross value of information is 315 − 72 = €243k; net value is 243 − 45 = €198k. Measurement is economically justified but does not guarantee a perfect answer. Run the test before renewal and decide on the lower contribution bound, not attributed ROAS.
Limits:
Information value depends on assumptions about losses and probabilities. A local experiment may not transport nationally. Interference and leakage weaken the counterfactual.