The Grandmother Test

Which channel actually pays?

Add up what your ad platforms claim and you will get more sales than you actually made — often 30–60% more. That is not fraud, it is arithmetic: each platform counts every sale it touched, so one purchase gets claimed three times. The only honest denominator is the orders in your billing system.

Gergana Tyaneva · 21 September 2026 · 12 years in product and marketing analytics

The grandmother version

Imagine you buy a coat. On the way to the shop you passed a billboard, a friend mentioned the brand, and a flyer came through your door that morning.

Afterwards, all three send you a bill for the sale. Each one honestly believes it caused the purchase. Add up their claims and three coats were sold. You bought one.

So don't ask each of them how much credit they deserve — they will all say a lot. Count the coats that actually left the shop, then work out how much more everyone claimed than was really there.

Why the numbers overlap

Every ad platform is measured on its own contribution and configured, by default, to claim generously: long click windows, view-through credit, and its own definition of what counts as an assisted conversion. None of them can see the others.

So the overlap is structural. A platform reporting well is supposed to claim a sale it touched. The error is yours only if you add those claims together and believe the total.

What the platforms claimedGoogle Ads620Meta540Email380Affiliate200What actually happenedPaid orders12421740 claimed ÷ 1242 real = 1.40× overclaim

The number worth building

The overclaim ratio is the single most useful figure you can put in front of a marketing meeting, and it takes an afternoon.

It converts an argument about attribution models into one observable fact: for every real sale, the platforms collectively claimed N. Once everyone has seen that number, budget conversations get shorter.

How to calculate your overclaim ratio

  1. Pull completed, paid orders for one full month from billing. Not from an analytics tool. Call this n.
  2. Pull each ad platform's reported conversions for the same month, on the same date basis.
  3. Add the platform numbers together. Call this c.
  4. Divide: c ÷ n. A result of 1.4 means the platforms collectively claimed 40% more sales than existed.
  5. Repeat for two more months. A stable ratio is a planning input; a moving one means something changed in setup or traffic mix.
  6. Report it as a count, never a bare percentage: "platforms claimed 1,740 sales against 1,242 real orders".

The short version

Last-touch attribution is a convention, not a truth. If you use it, label it as a convention — it is a tidy way to divide credit, not evidence of cause.

The only real evidence of causation is a holdout or a geo/time split: some people see the campaign, comparable people don't, and you measure the difference. Where no holdout exists, say the number is directional and resist dressing correlation as cause.

Or as your grandmother would have it: everyone who pointed at the shop will tell you they sent you there. Count the coats.

Which channel earns its budget is the core of Where the Money Comes From — €4,500, three weeks.

See prices →

More answers

Plumbing — Our numbers live in eight tools and no two agreeTime — Month-end reporting eats three days and nobody reads itMoney — We know people churn. We don't know who, when, or what it's worthThe real one — Someone asked me for a dashboard and I can't work out what for

One question before you read on. I'd like to count visits with Google Analytics, which sets cookies and sends your IP to Google. Nothing has loaded yet, and the site works exactly the same either way. What's collected.