What's the problem with attribution models?

Every attribution model — last-touch, first-touch, linear, time-decay, data-driven — encodes an opinion about how credit should flow. None is 'true.' The right approach is to pick a model that matches your buying cycle and decision speed, then triangulate with incrementality testing.

How do you make decisions when the data is noisy?

Three habits: (1) always estimate the noise floor before reading a result, (2) prefer multiple weak signals that agree over one strong signal that doesn't, (3) make reversible decisions fast and irreversible decisions slow. The Analytics chapter mini-game ('The Detective') trains exactly this.

What does the Analytics chapter mini-game (The Detective) involve?

The dashboard says one thing. The raw data says another. You investigate which is lying — usually a tracking bug, sampling artifact, or attribution change disguised as a trend. The game rewards skepticism and punishes dashboard worship.

What is incrementality testing?

Incrementality testing measures the lift a channel or campaign causes versus a holdout group that doesn't see it. It's the only attribution method that survives the death of third-party cookies and the rise of self-reported attribution.

Chapter 08 · Analytics — Growth Marketing

Growth / A Story

Ch. 08 · Analytics

Loading run_