Notes on measurement.
Short pieces on what we see in models, methods and the room when the deck lands.
The way you model paid search probably needs to change.
Paid and organic search are falling while online sales hold flat, and Google is putting ads into its AI answers. What that does to paid search in a Marketing Mix Model, and what I'd do about it.
OLS and Bayesian estimation, explained.
Two ways to estimate the same number, worked through by hand on a deliberately simple shampoo example. One uses only the data. The other blends the data with a starting belief – and that belief is the thing to ask about.
The four disciplines of Marketing Mix Modelling.
Most conversations about getting started in MMM jump straight to data and tools. The two disciplines that actually decide whether it works – people and process – are the ones nobody asks about first.
Marketing Mix Modelling is not an operational tool.
A model refreshed every week isn't giving you sharper strategic insight. It's presenting noise as precision, for decisions nobody could act on that fast anyway.
When Marketing Mix Modelling isn't for you.
Daily and weekly MMM is easy to buy now. It serves a need that almost no organisation actually has – even the ones that are exactly right for MMM.
The marketing ROI that was actually a price cut.
A model built from just sales and media data will credit the media for a price cut, and credit the brand for a rising market. The fix starts before a single number is collected.
The model can't see the ad.
A Marketing Mix Model can measure what a campaign did, and even the gap between two campaigns. It can't tell you why one worked and the other didn't – and the ad is often the reason.
Start small, then get bigger.
Vendors and consultants have every incentive to start big. The best ones resist that incentive – and so should you.
The last click isn't marketing.
Paid search that catches shoppers who have already decided what to buy looks like marketing's best channel. It probably belongs in the sales budget instead.
Three questions before you sign.
A short checklist for the moment the deck lands in your inbox and a decision is expected by Friday.
What an adstock is, and why the rate matters.
What an adstock is, why analysts use them, and why the adstock rate should come from your data – a short primer with an interactive chart.