Pull up any migration checklist and the Google side is covered: redirects and canonical tags, Search Console, crawl budget. What almost none of them cover is the layer where a growing share of your buyers now find you, and that layer breaks in its own ways. Engines keep citing URLs for weeks after the pages are gone. New stacks ship with crawler-blocking defaults nobody chose on purpose, and the entity familiarity a brand spent years building doesn't follow a 301.
This checklist runs in the order a migration does, from the month before the cut through cut day to the sixty days after.
Why AI visibility breaks differently
Start with one piece of evidence, because most of this checklist follows from it. One practitioner deliberately 404'd fifteen pages that carried active AI citations and watched what happened. Nine of them kept showing up in answers after they were dead, and the heavily cited ones hung on for three to four weeks.
I 404'd 15 pages with active AI citations — 9 kept showing up in answers, 2 for nearly a month
I deleted 15 pages from our site two months ago. All of them were showing up in AI answers at the time. I wanted to know what would happen next, and the answer turned out to be more complicated than "the citations disappear." Some context....
So AI citations are sticky, and the stickiness cuts both ways. For weeks after your cut, engines will keep sending buyers to your old URLs. Those URLs have to redirect, or each zombie citation turns into a 404 with your brand on it, served at the exact moment you're trying to look bigger and better.
Looking forward, the same stickiness means your new URLs won't get cited right away, so expect an overlap window where visibility seems to dip while caches refresh. The part I'd worry about most is that staleness hides breakage. A migration that quietly killed your AI visibility can look fine for a month, and by then everyone has stopped watching.
Two structural risks sit on top of that, and the first is pure plumbing. New infrastructure, whether a different CDN or a stricter security layer, routinely blocks AI crawlers by default, and nobody notices because Googlebot still gets through.
A same-domain replatform is only that plumbing job. A rebrand or domain change adds the work of starting a reputation, because the entity layer resets. Models that learned your old name from years of text don't carry that familiarity over to the new one, and no redirect map can pay that cost down.
Before the cut: T-minus 30 days
A month out, build the citation baseline. Run your fixed prompt panel, the 15-25 buyer questions that matter, across ChatGPT, Perplexity and Gemini, and log two columns separately: where you're named and which URLs get cited. Without that baseline you can't tell "the migration killed our visibility" apart from "we were never visible," and once the migration is done it's too late to find out.
Cross the panel results with your analytics and you have the protect-list, meaning every URL that carries active AI citations or AI-referred traffic. Those pages get first-class treatment in the redirect map, and wherever the migration allows, they move with their URLs preserved.
Two smaller jobs finish the prep. Snapshot the last month of AI crawler activity, which bots fetch what and how often, so you have a comparison set for confirming the new stack serves the same crawlers. Then spend thirty minutes exporting Search Console's Generative AI report and your GA4 AI channel data, plus whatever Bing Webmaster Tools shows, because migrations reset enough reporting surfaces that the before-picture is worth having.
See your first AI citation report in under 5 minutes.
No setup calls. No onboarding meetings. Connect your domain and see where AI mentions your brand right now.
During the cut
Cut day starts with the protect-list. Those redirects go live first and someone checks each one by hand, and everything else follows the normal 301 map. Classic checklists get the priority backwards here: SEO logic ranks redirects by search traffic, but an AI-cited page deserves equal rank even when its referred traffic looks small. A citation is a live recommendation being served to buyers right now, and a 404 behind it burns trust somewhere you can't see.
Before launch, audit the new stack's doors. Fetch key pages on staging as GPTBot and OAI-SearchBot, then as ClaudeBot and PerplexityBot. New CDNs and security layers tend to block first and ask never, and you want to find that block before the crawlers do.
If the replatform changed how pages render, check what the fetchers actually see by loading the content without JavaScript. AI fetchers parse server-rendered pages with semantic markup, which means a redesign that moved your pricing into a client-side component just took it out of every AI answer.
Then regenerate the machine surfaces. Submit the new sitemap everywhere, regenerate llms.txt if you keep one, and make sure the structured data carried over and still matches the visible content. On a domain change, keep the old domain alive and serving redirects for years rather than months, since the 404 experiment shows citations persisting for weeks at minimum and entity memory lasts longer.
After the cut: the sixty-day watch
For sixty days, rerun the panel every week with the same prompts and the same logging. A healthy migration shows a dip that recovers, with old URLs fading from answers and new ones appearing over several weeks. The dip that doesn't come back is what you're watching for, because it means something structural broke and the staleness window hid it.
Your 404 logs deserve some time too. Look for recurring paths that arrive with AI referrers or match your old URL structure. Engines keep citing dead URLs and now and then cite ones that never existed, and practitioners increasingly redirect those hallucinated paths to the closest real page so a dead end becomes a landing. You only know which URLs to catch if you're monitoring, and each catch recovers intent that would have hit a 404.
Checking crawler parity is quicker: compare AI crawler activity on the new stack with your T-30 snapshot. When a bot is missing, a door is closed somewhere, and the traffic-drop checklist runs that diagnosis in order. Finally, re-verify the reporting surfaces, which means GSC on the new property and Bing Webmaster Tools, plus updated GA4 channel group patterns if the domain changed.
Your competitors are building backlinks while you read this.
Organic outreach, social mentions, and citations from Editorial Circle member sites, with managed backlinks available as an add-on. Grow your domain authority without running the campaign yourself.
The recovery curve, annotated
Knowing the healthy shape ahead of time is what keeps week two calm.
| Days after the cut | What a healthy migration looks like |
|---|---|
| 1-7 | Answers still cite your old URLs heavily, the redirects absorb the clicks, and nothing looks different |
| 7-20 | Old URLs fade from some answers while new ones turn up in others, and your panel numbers wobble |
| 20-45 | New URLs carry a growing share of citations, engine by engine at different speeds; the slowest engines still serve stale answers past day 30 |
| 60 | The panel matches or beats the T-30 baseline |
That quiet first week is the staleness window doing you a favor. The wobble that follows is normal too, even though it's exactly the stretch where teams panic and start reverting things.
Failure looks different in kind, which is why the weekly rerun matters. If the curve dips and then flattens instead of recovering, or one engine goes dark while the others recover, you're looking at a structural break, such as a blocked crawler or a rendering gap, hiding behind what looked like normal cache lag.
Who owns this
Migrations are run by engineering and project managers, so the AI checklist dies unless it has a named owner inside the migration plan. Give your SEO or growth owner three dated deliverables: the T-30 baseline and protect-list, cut-day verification of redirects and crawler access, and the weekly panel through T+60. Then make the protect-list a formal input to the redirect ticket instead of a side-channel Slack message.
The condensed checklist
Print this half and pin it to the migration ticket.
| When | Task |
|---|---|
| T-30 | Fixed prompt panel baseline, citations and name-drops logged per engine |
| T-30 | Protect-list of every AI-cited URL, flagged in the redirect map |
| T-30 | AI crawler log snapshot, dashboard exports |
| Cut day | Protect-list redirects verified manually; staging fetched as all four major AI user agents |
| Cut day | Rendering check without JavaScript; sitemap and machine files regenerated |
| Cut day | Old domain redirect policy set to indefinite |
| Weekly to T+60 | Panel rerun, dip-and-recover watched per engine |
| Weekly to T+60 | 404 logs mined for zombie and hallucinated URLs, redirects added |
| T+14 and T+45 | Crawler parity check against baseline |
| T+60 | Full before-after readout against the T-30 baseline |
The special cases
How much of this you need depends on the kind of migration. A same-domain replatform that keeps its URLs is the gentle case: your work concentrates on crawler access and rendering, and citations mostly never notice. Restructuring URLs on the same domain sits in the middle, since the redirect map carries everything and your protect-list discipline decides how much sticks.
A domain change or rebrand is the hard case, because the entity layer resets however clean the plumbing is. Models learned your old name from the whole internet's text, and the new name starts from near zero. I'd budget real earned-media effort for the quarter after the cut, since engines recommend what the web already talks about, and the fastest entity rebuild is other people's pages using your new name.
So which migrations go wrong? Rarely the ones where a person owned this list. It's the ones where someone assumed the AI layer would follow the Google layer on its own, and that assumption fails because the AI layer has its own caches and crawlers, refreshes on a different cadence and keeps a separate memory.
Take the baseline before the cut and verify during it. Whether you find a broken door in week one or end up explaining a mystery in month three comes down to whether anyone is watching afterward, and weekly per-engine tracking makes the watching automatic.
You're getting AI traffic. But do you know where it comes from?
RankControl credits every visit to the assistant that sent it: ChatGPT, Perplexity, Claude, Gemini, Copilot, or Grok. Full source attribution, next to your Google traffic.
