Does GA4 count purchases from people who decline cookies?

Sometimes. In advanced consent mode the tag still sends cookieless pings, and GA4 reports those events once behavioral modeling is active. In basic mode the tag waits for consent, so a decliner's purchase never reaches GA4. Shopify still records every order either way.

What is behavioral modeling in GA4?

GA4's estimate of what visitors who declined analytics cookies did, trained on similar visitors who accepted. It fills in users and sessions, switches on by itself once a property qualifies, and shows only when the reporting identity is set to Blended.

Can I trust GA4 data after cookie consent?

Usually for trends, less for exact counts and channel credit. With advanced consent mode and enough traffic, GA4 models the visitors who decline. Without it, their visits go unreported and their purchases can land in Direct. Count your orders in Shopify.

By Joris van Huët, Founder & CEOUpdated 2026-10-017 min read

Usually for trends, less for exact counts or channel credit. If your banner runs consent mode in its advanced form and your traffic clears Google's thresholds, GA4 models the visitors who decline. If not, their visits go unreported and their purchases can land in Direct. Count orders in Shopify, and read GA4 for direction.

The usual answers come in two moods. One says consent mode and modeling fill the gap, so carry on. The other says half your visitors clicked no, so half your data is fiction. Google's own help pages land somewhere duller, and more useful.

Start with the panic. Google's Ads help says consented users are typically 2 to 5 times more likely to convert than unconsented users. In the example on that page, a 50% consent rate brings only a 19% drop in conversions. Losing half the cookies is not losing half the sales.

Then the reassurance, which has limits of its own. Modeling only switches on above a traffic threshold, and it estimates people, not paths. Most of this answer is about those limits.

What one store's data shows

One store's anonymised GA4 export, 1 January 2024 to 21 August 2026. It holds shares of revenue only: no ad spend, no order counts.

What the export shows	Share of revenue	Source cell
Direct, in last click, first click and touched views	57.7%	Channels sheet, Direct row
Journeys with 1 touch (0.5 days to buy)	79.5%	Journeys sheet, 1 touch row
Journeys with 2 to 3 touches (12.5 days to buy)	12.2%	Journeys sheet, 2-3 touches row

No row says who accepted cookies, because the export has no consent field. What it does show are the two places where a consent gap would hide.

First, Direct. It holds 57.7% of revenue in last click, first click and touched views alike (Channels sheet). Google's help on modeled key events says that when traffic is too thin to model, GA4 attributes those key events to Direct. So a big Direct row can hold buyers GA4 could not follow, not only people who typed your address.

Then the journeys. Journeys with one touch hold 79.5% of revenue and took 0.5 days to buy (Journeys sheet). Journeys with 2 to 3 touches took 12.5 days (Journeys sheet). A path that long exists only if GA4 kept recognising the same visitor the whole time. One visit with cookies declined breaks it, and the pieces look like shorter journeys.

So in the export, the 79.5% of one-touch revenue may be quick buying, or longer journeys cut into pieces. The export cannot say which. Your consent settings, and a comparison by country, can.

Modeling is real, but narrower than the reassurance suggests. Four limits decide how much of what you read was observed and how much was estimated.

It needs tags that load before consent. Google calls that the advanced setup. If your tags wait for a click on the banner, Google's Analytics help says you get no modeled data for the people who decline. In that basic setup, Google's Ads help adds, a decliner sends Google nothing at all, not even the consent status.

It needs volume. Google's eligibility rules ask for at least 1,000 events a day from visitors who declined, for at least 7 days. They also ask for at least 1,000 daily users who accepted, on at least 7 of the previous 28 days. Without enough consented traffic to learn from, Google says events from people who decline are not reported at all.

It estimates people, not actions. In reports, modeling fills in users, sessions and new users, while event counts such as page_view stay as observed. And because GA4 cannot tie a decliner's events together, first_visit and session_start fire on every page they load, so those counts run high.

Some places never see it. Audiences, segments with a sequence and retention reports leave modeled data out. So do user explorer, cohort and user lifetime explorations, predictive metrics and BigQuery export. Most exploration templates show consented users only.

Which ad a decliner saw. Google's Ads help says it cannot observe whether an unconsented conversion followed an ad interaction. Its models fill in the path, and Google says some real conversions go uncounted rather than over-predicted. So channel credit for those buyers is an estimate, or missing.

How big the estimate is. Reports mix observed and modeled users without marking which is which. The data-quality icon at the top of a report says whether estimates are included, not how many. Switching the reporting identity from Blended to Observed for a moment shows the gap.

What happened yesterday. Google says yesterday's estimates may not be ready, and key event credit can shift for up to 12 days. Read last week, not last night. The wider question of whether GA4 is accurate enough for budget calls has its own answer.

What to do this week

Read the data-quality icon for last month. Open Reports > Acquisition > Traffic acquisition for last month and check the icon at the top. Pass: it says Including estimated user data. Fail: Excluding estimated user data means your identity setting hides estimates; Estimated user data unavailable means decliners are simply missing.
Check the consent signals GA4 receives. In Admin, under Data collection and modification, open Consent settings and select your web stream. Pass: neither the advertising nor the behaviour signal section shows an alert. Fail: an alert appears, so fix the banner or its consent mode setting before you trust any EEA number.
Compare one month's orders in two countries. In Shopify, open Total sales by billing location. In GA4, put Country and Ecommerce purchases in a free-form exploration. Divide purchases by orders for a country where your banner shows and one where it does not. Pass: the two shares sit close together. Fail: the banner country's share is far lower, and that gap is roughly what consent costs your data there.

Check the homework. Your GA4 Attribution paths export already holds the evidence. Causality Engine reads that one file and shows what each channel caused next to what last-click gave it, in 1 to 2 minutes, for €99 once (excluding VAT), refundable within 30 days. Check the homework

Frequently asked questions

Does GA4 count purchases from people who decline cookies?
Sometimes. In advanced consent mode the tag still sends cookieless pings, and GA4 reports those events once behavioral modeling is active. In basic mode the tag waits for consent, so a decliner's purchase never reaches GA4. Shopify still records every order either way.
What is behavioral modeling in GA4?
GA4's estimate of what visitors who declined analytics cookies did, trained on similar visitors who accepted. It fills in users and sessions, switches on by itself once a property qualifies, and shows only when the reporting identity is set to Blended.
Will a cookie banner lower my Google Ads conversions?
Usually by less than your consent rate suggests. Google's Ads help says consented users are typically 2 to 5 times more likely to convert than unconsented ones. Once you qualify, Google Ads also models some of the missing conversions into its Conversions column.

Go deeper: Causal attribution, explained.

Sixty-second versions of these ideas: Causality Engine on YouTube Shorts.

Keep reading

Terms in this article

Browse the full glossary

What one store's data shows

Which GA4 numbers survive consent?

What can consent-hit data not tell you?

What to do this week

Related answers

Frequently asked questions

Keep reading

Terms in this article

Your platforms guess.We run the math.

Your platforms guess.
We run the math.