See what’s real and what’s guessed.
Every visit, the ones who said yes, and an estimate. Never blended.
Preview. The three-way split runs in the hub today. The estimate is one written rule, off until you switch it on.
Every sale, from your payments. A count, nothing else.
Estimates mixed in. Which part, it doesn’t say.
On every card, its kind: counted, observed, from payments, platform says, or estimated. An estimate is never added to a count, and never sent.
Most dashboards blend real and modelled numbers.
You can't tell what's real, so you can't defend it.
1,000
Events a day without consent, for at least 7 days, before GA4 will model the gap. Its reports then say they include estimated data, but never which number.
Source · Google Analytics Help, behavioural modelling for consent mode
How it works.
- 01Count everyoneEvery visit, on your own domain.
- 02See who said yesThe share a consent-only tool would have seen.
- 03Mark every estimateFrom a rule you can read. An estimate never passes as a count.
- Every visit
- 56,000 · every ad channel, counted
- Said yes
- 35,280 · consent recorded
- Estimated
- Off · marked wherever it shows when on
- A consent-only tool saw
- 63% of visits
- Estimate rule
- said-yes rate × declined visits
Illustrative · the hub’s screen is an example
Spend on numbers you can defend to clients and regulators.
Asked about real vs estimated
Do I still need a cookie banner?
For anything identifying or sent to an ad platform, yes. Counting alone doesn't need one in the UK. The EU hasn't changed its rule: there, you still need a yes first.
Why does the split matter?
It shows how much a consent-only tool misses. The gap between every visit and the visitors who said yes is traffic those reports never show you.
How is this different from Consent Mode?
Consent Mode models inside Google, by Google’s rules. This rule is written down, switched on by you, and the estimate is marked on every number it touches.
This explains the law in plain English. It isn’t legal advice.