A GA4 audit for a US industrial parts retailer on BigCommerce found that in one week, 20–27 Sep 2026, the store recorded 92 orders and GA4 only 49. We reconciled order by order, ruled out six causes and took the evidence to BigCommerce, which accepted it as a platform bug. By 19–26 Oct, GA4 captured about 85% of orders again; the platform fix is pending.
Sources: store order export and GA4 (order-confirmation views, purchase events), Sep – Oct 2026; test orders, Nov 2026.
The retailer sells industrial parts in the United States through a BigCommerce store. GA4 receives a purchase event when an order is placed, and Google Ads imports that purchase as its conversion. From 21 Sep 2026, part of the checkout orders stopped reaching GA4, while views of the order-confirmation page kept being recorded.
From 21 Sep 2026, about 50% of order-confirmation views produced a purchase event in GA4. From 1 to 20 Sep, about 89% had.
Google Ads imports the GA4 purchase as its conversion, so orders that came from ads were undercounted too.
Automated bidding optimizes toward the conversions it can see. With orders missing, it was learning from part of the picture.
Confirmation pages were still recorded in GA4, so the tag was loading. Something between the page and the purchase event was failing.
We started from the store’s own orders, not from GA4 reports. Each step narrowed the cause until only the platform was left. Then we gave BigCommerce a test it could repeat.
BigCommerce’s engineering fix was still pending at the time of writing.
Outcome43 of 92 orders missing; GA4 showed about 30% of that week’s store revenue.
OutcomeSix causes ruled out. The fault sat in how the purchase event was sent.
OutcomeBigCommerce accepted it as a platform bug and opened an engineering ticket.
OutcomeCapture back near 85% while the platform fix is pending.
OutcomePurchases now carry the customer’s hashed details, not the store’s.
Outcome43 of 92 orders missing; GA4 showed about 30% of that week’s store revenue.
OutcomeSix causes ruled out. The fault sat in how the purchase event was sent.
OutcomeBigCommerce accepted it as a platform bug and opened an engineering ticket.
OutcomeCapture back near 85% while the platform fix is pending.
OutcomePurchases now carry the customer’s hashed details, not the store’s.
We matched store orders to GA4 order by order in three weeks: before the loss, the worst week and after the escalation. Each bar is the share of that week's store orders found in GA4.
Sources: store order export, GA4. Order-by-order match.
A GA4 total can look plausible while orders are missing. Matching each store order to its purchase event shows which ones are gone and what they have in common.
If the confirmation page is recorded but the purchase is not, the tag is loading. That rules out blocking, consent and filters in one step and points to whatever sends the event.
Support teams act on cases they can reproduce. A written test-order procedure, with what should arrive and what did, turned a reporting gap into an engineering ticket.
A tracking setting can be switched on and still send the wrong person’s data. Only a real test order shows what the purchase event actually carries.
Something else? Email hello@semroi.net, or read our GA4 cross-domain tracking guide.
A GA4 audit, or Google Analytics audit, checks that what GA4 records matches what happened. For an online store, that means comparing GA4 purchases with the store’s own orders, checking values, filters, consent and tags, and confirming Google Ads imports the right conversion.
If the confirmation page is recorded but the purchase event is not, the tag is loading, so blocking and consent are unlikely causes. Look at whatever sends the event. Here it was a platform bug in BigCommerce GA4 tracking, accepted by BigCommerce, hitting desktop Chrome and Edge most.
Export the store’s orders for a period and match each order number against the transaction IDs in GA4. Orders with no match are missing. In this audit, 20–27 Sep 2026 had 92 store orders and 49 in GA4. Daily confirmation views against purchase events showed when the loss began.
When Google Ads imports its purchase conversion from GA4, every purchase GA4 misses is missing in Google Ads too, and automated bidding learns from partial data. Check GA4 against store orders first. Changes on the Google Ads side alone will not bring those conversions back.
Place a real test order and inspect the purchase event it sends. In this audit, the event sent the store’s own e-mail address as user data instead of the customer’s. After the fix, a second test order confirmed the purchase carries the customer’s details, hashed.
Share a week of store orders and read access to GA4. We will match them order by order and tell you what we find.
