Research on measurement
Research: measure before you attribute
This collection compares marketing-measurement methods through the decision they can support, their data requirements and their limits. A contribution attributed by a platform is not automatically a causal effect.
01 · Analyses
Compare the evidence, not the labels.
Each analysis states the question addressed, the data required, the kind of causality within reach and the decision the method can genuinely inform.
Attribution, incrementality or MMM: which method for which decision?
Choose a measurement approach by estimand, data, bias and decision horizon; do not mistake attributed contribution for causal effect.
8 August 2026 →Scientific analysisReach, frequency and media pressure: no universal threshold
Set reach and frequency with an explicit budget curve, measurement conditions, uncertainty and reversible test plan—not a magic number.
11 August 2026 →02 · Evidence contract
Four checks before concluding.
- 01Name the question
Describing a journey, estimating a causal effect, explaining a variation or arbitrating the mix do not call for the same method.
- 02Audit the data
Document the grain, the length of history, the measurement breaks and the variables that are unavailable.
- 03State the counterfactual
Set out what makes it possible to distinguish an observed exposure from an effect actually caused.
- 04Triangulate the decision
Compare attribution, experimentation, MMM and business knowledge before moving a budget.