Writing

Essays on causal marketing measurement

Long-form, practical, and written for the people who have to defend the budget.

Series · 8 parts
MMM in practice
From data to a budget decision, one honest step at a time.
Series · 4 parts
Experiments that work
Designing, running and reading geo tests without fooling yourself.

MMM in practice · 31 May 2026 · 3 min read

Your attribution report can't answer the budget question

Why last-click and platform reporting measure activity, not cause, and what to ask instead.

19 essays

MMM · 14 Jun 2026 · 2 min read
A number without a range is a guess you've stopped admitting
Why every marketing estimate needs its range, and how to read one in a meeting.
MMM in practice · 9 Aug 2026 · 2 min read
Channels that move together can't be told apart
Collinearity in plain English, and three ways out of it.
Data · 13 Sept 2026 · 2 min read
Cookies, consent and iOS: what measurement still works
Privacy changes hurt attribution far more than mix modelling and experiments.
Experiments that work · 28 Jun 2026 · 3 min read
Design the test before you spend the budget
Power, placebo checks and comparison regions, explained for marketers.
Experiments that work · 23 Aug 2026 · 2 min read
Don't peek: why early test results lie
Early results swing. Locking the readout protects the decision.
Decisions · 27 Sept 2026 · 2 min read
Explaining causal measurement to your CFO
The four questions finance asks, and answers they'll accept.
MMM in practice · 21 Jun 2026 · 3 min read
Five checks before you trust a mix model
A field checklist for the people who have to sign off the budget.
Decisions · 11 Oct 2026 · 2 min read
From reports to decisions: five stages of measurement maturity
A self-assessment, from dashboards to calibrated decisions.
Experiments that work · 6 Sept 2026 · 2 min read
Holdout tests when the stakes are high
Running geo holdouts when brand risk, compliance and senior attention are real constraints.
MMM in practice · 30 Aug 2026 · 2 min read
How a geo test should change your mix model
Calibration: making models and experiments check each other.
MMM in practice · 19 Jul 2026 · 2 min read
If your spend never changes, no model can measure it
Why always-on, flat budgets make channels invisible to measurement, and the simple fix.
Decisions · 20 Sept 2026 · 2 min read
Plan next quarter with the ranges left in
Forecasting honestly: carryover, assumptions and ranges that widen the further you look.
Attribution · 7 Jun 2026 · 3 min read
Platforms grade their own homework
Why platform-reported conversions overstate impact, and the test that settles it.
Data · 4 Oct 2026 · 2 min read
Server-side tracking fixes some things, not everything
What moving tags to a server genuinely improves, and what it can't.
Data · 26 Jul 2026 · 2 min read
The weekly dataset nobody wants to fund
The unglamorous work that decides whether any model means anything.
MMM in practice · 12 Jul 2026 · 2 min read
TV doesn't stop working when the ad stops
Carryover, and why weekly reporting undervalues brand channels.
Experiments that work · 16 Aug 2026 · 2 min read
What a placebo test tells you that a p-value can't
Checking your result against places where nothing happened.
MMM in practice · 2 Aug 2026 · 2 min read
When two models disagree, that's the finding
What running two modelling engines on the same data tells you that one never will.
MMM in practice · 5 Jul 2026 · 2 min read
Why the next pound is never worth the last
Saturation explained with one curve, and what it means for your budget.