No one can give you this number, but you can calculate it in five minutes: divide the purchases GA4 sees by the orders recorded in Shopify.
What's missing comes from declined cookies and blockers. In the brands we've seen, the most common answer comes down to two words: about 60%, and nobody saw the rest.
The benchmark
Take a full month.
At one store the authors advise, with a little over a thousand orders a month, GA4 counted 846 purchases for 1,400 Shopify orders: 60% coverage.
Average order value was identical in both tools, within one euro. GA4 wasn't wrong about value. It saw fewer orders and more visits, which cut the conversion rate in half (0.89% versus 1.78%).
The cause was a Google consent mode that had never been turned on. Run the calculation again by device: the gap is rarely the same on mobile and desktop.
The rule
Track this coverage rate every month. If it's stable, your GA4 trends hold up. If it suddenly shifts, look for what changed in your banner or your tagging before you comment on your performance.
The steps
- Overall coverage. Calculate it, then break it down by device and by browser.
- The logbook. Record every change to the consent banner.
- Modeling. Check whether it's active on your property. Google turns it on only above specific thresholds: at least 1,000 events per day without consent for seven days, at least 1,000 users per day with consent on seven of the last twenty-eight days, and the reporting identity set to "Blended." Below that, GA4 fills in nothing.
- The reference. Use Shopify for totals.
- The line for leadership:
"GA4 sees six orders out of ten. We read its trends, never its totals."
The trap
The loss isn't spread randomly. Some audiences decline more than others, and some browsers block more.
So GA4 undercounts some channels more than others, and your channel comparison is skewed with no sign that anything is wrong.
And consent doesn't cripple only GA4. At a fashion brand, Shopify Analytics tied a device to just 285 of 568 orders: orders placed without analytics consent count in the total, but lose their device, source, and country. Your mobile versus desktop split, even on the Shopify side, describes only the consenting half of your customers.
As for making it harder to decline in order to lift your measurement: the CNIL has required since 2021 that declining be as easy as accepting.
