The "optimise your container" banner in Google Tag Manager takes a few clicks. Checking it worked takes an hour. Here's a checklist, in order, so the speed gain doesn't cost you a week of broken GA4 or Google Ads data.

If you want the background first, start with what the new tagging experience changes.

Before you click.

  1. List your Google tags. Write down every Google tag, GA4 configuration and Google Ads or Floodlight tag in the container, with its ID and settings: fields to set, user properties, server container URL, and anything unusual. That list is what you check the result against.
  2. Check who can see them. Optimising gives everyone with access to a linked Google tag Read access to the container by default. Look at who has access to each GA4 property and Google Ads account first, and remove anyone who shouldn't be able to read your tagging.
  3. Find stray config calls. Open Preview and look for a gtag.config event. A hard-coded config call outside the container will keep sending alongside the optimised container, so deal with it first.
  4. Note your consent setup. Write down how consent defaults are set today, by the consent platform's template or by a tag, and what fires before anyone clicks the banner. You'll test the same thing afterwards.
  5. Start a clean workspace. Do the optimisation on its own, with nothing else unpublished, so it can be published and rolled back as one version.

During the optimisation.

  1. Untick what you don't own. The flow lists every Google tag it finds. Untick any you don't administer, such as an agency's Floodlight. They stay as legacy tags and carry on as before.
  2. Decide on Read access. You'll be asked whether to give container access to people with access to each Google tag. If you did the check above, saying yes is fine.
  3. Choose the settings carefully. If your Google tags have different settings, you pick one tag's settings as the blueprint or set them from scratch. This is where an old setting gets lost, so compare against your list.
  4. Read the Before and After. Open the details. Linked tags are converted to the new destination tags and shown as modified. Check each one's ID and overrides.
  5. Add to the workspace, not live. Nothing is published at this point.

Test in Preview.

  1. Check the Settings tab. Preview now shows the container settings in use. Make sure consent defaults and data collection settings match what you intended.
  2. Test with consent refused. Load a page, refuse cookies, and look at the Hits sent tab. Deferred hits are flagged, and you should see nothing sent that wasn't sent before.
  3. Then accept. Grant consent and check that held hits are sent and that later page views and events reach every destination.
  4. Watch for delayed hits. A "hits being delayed" warning means a tag fired before its destination had loaded. Check it isn't your page view or a purchase.
  5. Run a conversion. Complete a test purchase or form, and check the GA4 event and the Google Ads conversion both arrive once, with the right values.
  6. Check the speed gain. In the browser's network panel, the separate gtag.js requests for your linked destinations should be gone.

After publishing.

  1. Name the version clearly, and note the date in GA4, so a change in the figures can be traced back to it.
  2. Compare the next few days with the week before: page views, key events, Google Ads conversions and consent rates. A sudden drop is a reason to roll back first and investigate after.
  3. Rolling back is simple: publish the previous version. Access links aren't versioned, so review them separately.

When to wait.

There's no deadline. Wait if the container sends to several destinations with competing settings you can't yet resolve, if a key destination belongs to someone you can't reach, or if your consent setup depends on rules for non-Google tags that the new data transmission setting doesn't cover.

On a container like that, add destinations one at a time, testing between each. Google gives no warning when container settings and a destination's own settings conflict, so the testing is what catches it.

Marc can do it for you as part of his Tag Manager work, and a first look at your container is part of the GA4 mini audit.

Sources: Google's optimisation guide and Simo Ahava's walkthrough.