Library·SOP·Analytics
Add one Google tag through Google Tag Manager, test it in Preview, and confirm that GA4 receives a normal page view after publishing.
Last reviewed: · Editorial review
Clarity, structure and references reviewed in this version. This is not a test in a live account.
Add one Google tag through Google Tag Manager, test it in Preview, and confirm that GA4 receives a normal page view after publishing.
This SOP covers:
This does not cover:
Open Google Analytics. Select your property, then open Admin → Data streams and choose the web stream for your website. Check its website URL and that Enhanced measurement is off. Create or open a note named Measurement handoff note for this setup. Copy the stream's Measurement ID, which starts with G-, into that note.
Open Google Tag Manager and select the installed Web container. Match its GTM- ID to your website’s container ID. Review Workspace Changes and open Tags to check for an existing Google tag or GA4 configuration. If old tags are retained, check that they are marked Paused.
Confirm that the published container is the starting point and that the live pages do not already run an active Google tag.
Open Tag Assistant, connect to your first page URL, and follow the website's normal consent flow for an analytics-enabled visit.
Select your GTM container. Under Container Version, open the version selector, choose Debug for the Live environment, and click Reload page. Confirm that Container Version reads Live.
Inspect the detected tags on the first page, then open the second page in the same connected session. Confirm that the expected GTM container loads on both and that no active Google tag or Analytics destination appears.
Click Stop debugging when the baseline check is finished.
In your GTM workspace, choose Tags → New. Name the tag Google tag – GA4. In Tag Configuration, select Google tag and paste the G- measurement ID into Tag ID. Under Triggering, choose Initialization – All Pages, then save. Reopen the saved tag and check that its Tag ID matches step 1.1 and its trigger is Initialization for all pages.
Check the property’s active data filters before testing. Use an approved test network that is not excluded by an internal-traffic filter for the normal Realtime check; do not disable production exclusions to make a test pass. Record the network and filter state with the evidence.
Test one Google tag on two full page loads in Preview. Tag Assistant shows what the draft sent; the normal GA4 Realtime receipt comes later.
Click Preview in GTM. Enter the first page URL, click Connect, and return to Tag Assistant to click Continue. Follow the site's normal consent flow.
Select the page's Initialization event and check that your Google tag fired once. Select the Google tag's destination and inspect its page_view event, GA4's record of a page load. Match the G- ID and page_location to your stream and first page.
Open the second page in the same connected session and repeat the Initialization, destination, G- ID, page_location, and page_view check.
In the Measurement handoff note, record the two checked page URLs and the one page_view result for each full page load.
Before publishing, record the owner’s approval for the tested change and the currently live GTM version in Measurement handoff note. If live checks fail because of this release, have the authorized publisher restore that recorded version and repeat the checks. Review intervening releases before restoring a version so another person’s changes are not lost.
In GTM, click Submit and choose Publish and Create Version. Review the changes: only the Google tag you tested should be included. Add a version name and description, then click Publish. Open the published version and check that the saved Google tag still has your G- ID and the all-page Initialization trigger. Record the version number in Measurement handoff note.
Check that the published container sends page views without relying on the workspace draft.
In GTM Versions, confirm that the version number from step 4.1 is marked Live.
End the Preview session with Stop debugging and close the previewed website window. Start a fresh connection from Tag Assistant to your first page. Select your GTM container, open the selector under Container Version, choose Debug for the Live environment, and click Reload page. Confirm that Container Version reads Live.
Follow the normal consent flow. Select the Google tag destination on the first page and check one page_view to your G- ID. Open the second page in the same connected session and repeat the check.
Click Stop debugging and close the connected website window when the Live check is finished.
Confirm that GA4 received a normal visit outside debugging, using Realtime.
In GA4, open Reports → Realtime overview and note the current page_view count and the time in Measurement handoff note before opening the test pages.
Open your two page URLs in a normal tab and follow the site's consent flow. Use a unique, non-personal query label on each URL to distinguish this visit: append ?check= followed by a short label you choose, or &check= if the URL already contains a question mark. Confirm that both pages load normally and keep each label in the address bar.
Return to Realtime and refresh it. Under Event count by Event name, click page_view, then page_location. Find both complete labeled URLs and confirm one new event for each. Realtime is the receipt for this normal visit; DebugView is for debug visits and cannot replace this check. Allow up to 15 minutes for initial collection and refresh Realtime before deciding.
If the labels disappear or the pages stop working, use two identifiable page URLs and match their new events in View user snapshot instead. If both pages cannot be matched, stop and check the property, destination, and normal consent flow without bypassing privacy controls.
In GA4, return to Admin → Data streams, open your web stream, and choose Configure tag settings. Read Tag quality and open View diagnostics if issues are listed. In Measurement handoff note, record the status and any collection issues. Treat an issue reporting blocked, missing, or stopped data collection as an incomplete installation. Recommendations alone do not invalidate the result; a pending status does not replace the event checks.