GA4 shows roughly twice the purchases and events your store actually gets. Shopify or WooCommerce admin says 40 orders, GA4 says 78. In most stores I audit, the cause is simple: GA4 is installed twice - once by a plugin (Site Kit, a Shopify app, a theme field) and once through Google Tag Manager. Every event fires down both pipes.
This is a fifteen-minute diagnosis and usually a five-minute fix. Here is how I confirm it, why it wrecks more than the purchase count, and how to clean it up without losing data.
Open your store in an incognito window. Open DevTools (F12 or Cmd+Option+I), go to the Network tab, type collect in the filter box, reload the page. A healthy install produces one request containing en=page_view per measurement ID. If you see two requests with the same tid=G-XXXXXXX and en=page_view, you are double-tagged:
https://region1.google-analytics.com/g/collect?v=2&tid=G-ABC1234&...&en=page_view https://www.google-analytics.com/g/collect?v=2&tid=G-ABC1234&...&en=page_view
Three more checks that settle it:
You can also view page source and search for both gtag/js?id=G- and googletagmanager.com/gtm.js. If both are present and the GTM container also holds a GA4 tag, that is your duplicate.
Each GA4 config on the page dispatches every event once. Two configs for the same measurement ID means two hits per event: page_view, add_to_cart, purchase, all of it.
GA4 has one built-in guard for purchases: it drops a purchase whose transaction_id matches one it already recorded from the same user. That guard only works when both installs send an identical ID. Two independent installs rarely do. The plugin sends the raw order ID, GTM sends a prefixed order number, or one omits the field entirely. Both purchases count. Revenue doubles.
Sessions inflate for a duller reason: the two installs initialize at different moments with different settings, and redirects through payment gateways or consent updates can restart a session under one config but not the other. Even when session counts hold, event counts double, so every ratio built on the reports - conversion rate, revenue per session, AOV per channel - is computed from corrupted inputs.
The same order now exists as two purchase events with different timing and session context. One fires the moment the thank-you page loads, the other after a consent update or a gateway redirect. GA4 attributes each copy independently, so a single order can show under two channels, or the late copy lands in Direct.
If you import GA4 key events into Google Ads, Ads receives doubled conversions. Smart Bidding optimizes toward a ROAS that does not exist and spends accordingly. And because GA4 no longer reconciles with your store admin, Meta, or your bank, every budget argument turns into a guess.
Pick one install and kill the other. My default: keep GTM, remove the plugin's Analytics output. GTM gives you one place to manage GA4, Google Ads, Meta and consent, and it upgrades cleanly to server-side tagging later.
Do not delete the GA4 property or the data stream. Your history lives in the property. You are removing one of two loaders, nothing else.
Purge page cache and CDN first (WP Rocket, LiteSpeed, Cloudflare), otherwise the old snippet keeps serving. Then, in a fresh incognito window:
Next day, compare GA4 purchases against your store admin. GA4 at or slightly below the admin count is healthy - browser-side tracking always loses a share to ad blockers and declined consent. GA4 above admin means something still fires twice. Note the fix date: GA4 does not rewrite history, so reports before that date stay doubled and any period comparison has to start after it.
No. The GA4 property and its history live in your Google account, not in the plugin. Disconnecting only removes the tracking snippet Site Kit injects, and Search Console plus the other Site Kit modules keep working.
It is the most common cause, and a ratio near 2x makes it almost certain. A healthy browser-side setup reports slightly fewer purchases than the store admin, rarely more - cancelled or failed orders can put GA4 slightly above the completed-order count. A gap approaching 2x deserves the network tab check.
It does, when the same user sends two purchase events with an identical transaction_id. Two independent installs rarely send identical IDs - one uses the raw order ID, the other a prefixed order number, or one omits the field - so the guard does not save you.
Keep GTM if you run paid traffic, consent mode, or plan server-side tagging, since it is the one place to manage all tags. A plugin-only install is fine for a simple content site with no ad spend. Never run both for the same measurement ID.
I find and fix duplicate GA4 installs, broken purchase tracking and attribution problems weekly, mostly for stores running paid traffic. If you want this checked and fixed on your store, email work@rytisbalys.com with your store URL.
work@rytisbalys.com