Skip to content
Keyword Adv
← All field notes

22 Jul 2026 · 3 min read

Nobody tells the ads guy about the migration

Site replatforms happen in a vacuum. Then the tags die, the feed breaks, the URLs change, and the ad account finds out at 2am. A war story and a checklist.

A client replatformed their store this year. New site, new theme, new checkout. The web team did their job, the launch went fine, everyone celebrated. Nobody told me.

Here is what the ad account experienced that night. Every conversion tag stopped firing, because the new theme did not carry the Tag Manager container. The product feed broke, because the URLs changed and Merchant Center started disapproving items in batches. The Shopping campaigns kept spending against a feed that was rotting underneath them. Smart bidding, suddenly starved of conversion signal, started flailing. By the time anyone looked, the account had spent days optimising toward silence.

None of this is rare. I have seen versions of it on migrations big and small, and the cause is always organisational, not technical. The site rebuild lives with a developer or a web agency. The ad account lives with me. The client assumes someone connected those dots. Nobody connected those dots.

The brutal part is the asymmetry. The migration takes months of planning. The damage takes hours. And the recovery takes weeks, because you are not just fixing tags, you are refeeding a machine that learned garbage. Performance Max does not have an undo button. Whatever it learned during the blackout, it keeps.

So here is the checklist I now force into every engagement, and if you run ads for any business that might ever touch its website, steal it.

Before launch: get the migration date in writing. Confirm who owns moving the Tag Manager container and the GA4 config to the new site, with a name attached. Export the full URL map, old to new, because redirects that work for SEO can still break your feed and your sitelinks. Check whether the checkout flow changes, because that is where purchase events live and die.

Launch day: fire a test transaction and watch it land in GA4 and Google Ads in real time. Check Merchant Center for disapprovals within hours, not days. Keep budgets steady but watch conversion lag like a hawk. A quiet afternoon can be normal. A quiet 48 hours is a siren.

After launch: reconcile platform revenue against GA4 for the first two weeks. Any gap over ten percent means something is misfiring. Recheck every conversion action, including the boring ones like phone clicks that nobody remembers exist. And if the blackout lasted more than a couple of days, consider whether your bidding strategy needs a manual floor under it while the signal rebuilds.

One more thing that costs nothing and saves everything: put a line in your client agreement that says website changes get flagged to whoever runs the ads, before they ship. Not because you want to approve anything. Because the ad account is a system that feeds on the website, and you cannot keep a system healthy when its food source changes without warning.

The developer is not the villain here. Their job was the website and they did it. The gap is that in most businesses, nobody's job is the seam between the website and the money that traffic costs. Make it someone's job. Ideally before the next launch party.

Frank online