What we found checking 975 Shopify stores.
On 14 November 2026 we read the public pages of 975 Shopify stores and looked for visible tracking set-up issues. This page sets out exactly how the check worked, the counts, and the limits. These are screening observations, not confirmed faults.
Study summary and date
We examined publicly accessible pages from 975 Shopify stores on 14 November 2026. Our checks identified at least one potential tracking issue in 389 stores. These are screening observations, not confirmed checkout failures. Public pages cannot establish whether a purchase event fires correctly, and some implementations are not visible to this type of check.
975
Shopify stores checked, all on 14 November 2026.
389
Stores with at least one visible potential issue. 40% of this selected sample.
Where the sample came from
This was a selected sample of 975 Shopify stores, including London shops sourced from OpenStreetMap and other stores. It was not a random survey of all UK ecommerce websites.
No store is named on this page or anywhere else on this site.
What public pages can reveal
The check is a script we wrote. It sees what any visitor's browser receives, and nothing more. No logins, no forms, no purchases. For each store it:
- Read the homepage HTML.
- Found one product through the store's public product list and read that product page. This worked for 926 of the 975 stores; for the rest, only the homepage was read.
- Read Shopify's list of configured pixels, which appears in the page source. It shows app pixels, such as those from the Google & YouTube app and the Facebook & Instagram app, with their configured IDs, and the number of custom pixels.
- Where a Google Tag Manager container was found in the theme, read that container's public
gtm.jsfile to see which Google tags it holds. Up to three containers per store were read; a container was read for 236 stores.
To avoid counting stray mentions, an ID only counted as loaded when there was a loader or configuration call for it, for example a gtag/js script, a gtag('config', …) call, a Universal Analytics ga('create', …) call or a Meta fbq('init', …) call. Inside a Tag Manager container, a GA4 ID only counted when the container had a live Google tag and exactly one GA4 ID, because containers often hold lookup tables of IDs for different country sites. Paused tags were ignored.
During the build we found and removed three kinds of false positive: a theme sending events to a GA4 ID it never loads, an ID mentioned in data-layer configuration without being loaded, and a leftover Universal Analytics settings variable in a container with no Universal Analytics tag.
Counts by indicator
| Indicator | Rule used | Stores |
|---|---|---|
| No GA4 or GTM found in the public-page check | No GA4 ID anywhere in the pages read, no Tag Manager container, no custom pixel, and Shopify's pixel list present | 196 |
| Same GA4 ID loaded twice | One GA4 ID loaded by two or more of: theme code, Tag Manager, the Google & YouTube app | 92 |
| GTM observed in theme without a detected Shopify custom pixel | Tag Manager in theme code, no custom pixel, and no app pixel referencing that container | 74 |
| Universal Analytics setup detected | A Universal Analytics ID loaded in the page, or a live Universal Analytics tag in the container | 46 |
| Multiple GA4 IDs found | Two or more different GA4 IDs across theme, Tag Manager and the Google app | 38 |
| Meta pixel problems detected | See the breakdown below | 25 |
The 25 Meta pixel observations break down as: two different pixel IDs loaded, one in the theme and one by the Facebook & Instagram app (15); the same pixel loaded by both the theme and the app (5); a pixel in the theme only, with no Facebook & Instagram app pixel (5).
The check also looked for a Google & YouTube app configured without a purchase event. It found none.
Why indicators overlap
One store can show several indicators. A store with GA4 loaded twice may also have a second GA4 ID and old Universal Analytics code. Of the 389 stores, 321 showed one indicator type, 55 showed two, 12 showed three and 1 showed four.
So the rows above can't be added up. Their total is more than 389 because of this overlap.
What the check cannot establish
- Whether purchases are tracked. Checkout and order status pages are not public. A store flagged "GTM in theme without a custom pixel" may track purchases another way. The absence of a detected custom pixel does not prove purchase tracking is broken.
- Whether duplicate tags cause duplicate conversions. A GA4 ID loaded twice can count events twice, but GA4 and the tags involved may prevent it, for example through transaction ID deduplication of purchases. It needs a test to know.
- Anything loaded later by scripts. The check reads HTML. It doesn't run JavaScript, so tags added at run time by apps or consent tools can be missed. "No GA4 or GTM found" means we didn't see one, not that none exists.
- Server-side tracking. Conversions API, server-side Tag Manager and other server routes are invisible from public pages.
- Consent behaviour. We didn't interact with any cookie banner.
- Revenue impact. None of these counts show that any store lost money.
Limitations and sample bias
- The sample was selected, not random. It leans towards London shops listed in OpenStreetMap, which tend to be physical businesses with an online store. It doesn't represent all UK Shopify stores, or all ecommerce.
- The check ran once, on one day. Stores change their set-up all the time.
- Only one product page per store was read, and only the homepage where that failed.
- The rules were written to avoid false positives, which means the check will also miss some real issues.
The 40% figure describes this sample only. Please don't quote it as a rate for UK ecommerce.
Practical takeaways for merchants
- Check how many ways GA4 is installed. Theme code, a tag manager and a platform app can each load it. One method per ID is easier to keep right. See duplicate GA4 tags.
- Remove Universal Analytics code. Universal Analytics stopped processing data on 1 July 2023. See Universal Analytics code.
- Know where checkout tracking comes from. On Shopify, theme code doesn't run in checkout. Shopify's help on the Thank you and Order status upgrade says tracking there runs through app pixels and custom pixels. See Shopify GTM and custom pixels.
- Check your Meta pixel IDs. Two different pixels split your data. See Meta Pixel not firing.
- Test with an order. Only a test order proves purchase tracking. Our Shopify tracking checklist walks through it.
Updated methodology notes
- 14 November 2026: first run of the check on 975 stores, after the three false-positive rules above were added. All counts on this page come from that run.
- Any later run will be published with its own date and counts, and won't overwrite these figures.
Check your own store
A free check is a human review of your store's public pages by Mikael, with a plain note of what he could and couldn't see. To confirm purchase tracking with access and a test order, see Shopify tracking.
Sources
- Google Analytics Help: GA4 has replaced Universal Analytics
- Google Analytics Help: Minimize duplicate key events with transaction IDs
- Shopify Help Center: Upgrading the Thank you and Order status pages
- Shopify Help Center: Create a Google Tag Manager custom pixel
Method and counts reviewed against the check's code and output files on 17 November 2026.
Not sure what's wrong?
Send us your store address. Mikael will look at the public pages and tell you what he can see, and what he can't.