Chapter 10
Replacing old tracking, testing and sign-off
 "Zoran Artmagic")
Table of contents
The most dangerous moment is the day after go-live. Suddenly the numbers in GA4 rise sharply and everyone is pleased. Often the reason is not better tracking but double tracking: the old tracking simply keeps running.
How do you switch off the old tracking cleanly?
Use the inventory: The list of all tags, pixels and apps from the pre-analysis is the starting point.
Rebuild Tag Manager: The tags in the web container now send to the server. Direct connections to platforms that now run via the server are removed or reduced to the role they keep in the hybrid setup.
Review store extensions and apps: Old tracking plugins in Shopware and duplicate app pixels in Shopify are deactivated. The Tag Manager extension stays; it establishes the connection to the server.
Look for hard-coded scripts: Pixels in the theme, in template extensions or in old Shopify settings are the most common leftovers.
How do you test the new setup?
Testing means triggering real events and checking at every point whether they arrive.
Test purchases with different payment methods, including redirects via PayPal or Klarna.
Forms: fill in and submit them, for example contact, quote request or newsletter.
Product views, cart and checkout steps: click through them.
With and without consent: without consent, no marketing tags may fire.
In different browsers, especially Safari on the iPhone.
The tools for this are the preview mode of the web and server containers, DebugView in GA4, test events in Meta and the diagnostics in Google Ads. With TAGGRS, container validation is added.
How do you know the switchover has worked?
Not by the numbers being as high as possible, but by them being consistent. After two to four weeks, the purchases measured in GA4 should be closer to the actual orders than before, but never above them. If they are above, you are counting twice. The ad platforms should report more attributed conversions, and the share of measured Safari users should rise relative to Chrome. Compare like-for-like periods where possible, without promotional weeks, Black Week or seasonal peaks on only one side.
How long should old and new tracking run in parallel?
For comparison purposes, it can make sense to run the new setup in a separate GA4 property at first. That does not apply to the ad platforms: each conversion may only arrive there once. The old route is therefore switched off in the same step as the new one is activated.
Sign-off checklist
All old tracking plugins, app pixels and hard-coded scripts deactivated
Test purchases carried out with all relevant payment methods
Forms, cart and checkout steps tested
Tested that no marketing tags fire without consent
The purchase arrives exactly once each in GA4, Google Ads and Meta
After two weeks, measured purchases are not above actual orders
Frequently asked questions
Probably not. Such a sharp rise usually points to double measurement because the old tracking is still running. Compare the measured purchases with the actual orders in your store.
The technical tests take hours; observing the numbers takes two to four weeks. Only then can you reliably judge whether the measurement is consistent.