A website migration changes URLs, templates, and often entire content sections at once — and AI answer engines don't refresh their index on your schedule. Track AI visibility after a website migration by locking a pre-migration baseline of citations and prompts, then re-running that exact prompt set against ChatGPT, Gemini, Perplexity, Claude, and Microsoft Copilot at day 7, day 30, and day 60 post-launch. The hidden cost most teams miss: AI engines work off cached crawls that can lag weeks behind your live site, so a broken redirect or a stripped-down migrated page can quietly vanish from AI answers a month after the cutover, long after the SEO team has declared the migration a success.
- Track ai visibility after a website migration by rerunning one fixed prompt set at day 7, 30 and 60 post-launch.
- AI engines cache old crawls, so citation loss can surface weeks after the domain or URL change.
- A pre-migration AI Visibility Score baseline is the only proof a post-launch drop is real and not noise.
- Redirect chains and stripped schema markup are the two most common causes of lost AI citations in 2026.
Why this matters
Organic traffic dashboards recover fast after a clean migration — rankings settle within 2-4 weeks in most cases. AI citations behave differently, because large language models don't re-crawl and re-index on the same cycle as Google Search.
A brand can hold its keyword rankings within days of a migration and still disappear from ChatGPT and Perplexity answers for a full quarter, because the model's training or retrieval snapshot still points at a URL structure that no longer resolves cleanly. Nobody notices until a prospect says "I asked ChatGPT and it recommended your competitor instead."
How do you track AI visibility after a website migration?
The process has four steps, and skipping the baseline step is the single most common reason teams can't prove what the migration cost them.
- Lock a baseline 1-2 weeks before cutover. Run your full prompt set (the questions buyers actually ask AI assistants about your category) against every platform you care about and record which pages get cited, not just whether the brand is mentioned.
- Freeze the prompt set. Don't rewrite or expand it mid-migration — the point is a controlled comparison, not a fresh audit.
- Re-run the identical prompts at day 7, day 30 and day 60 after launch. Early checks catch crawl and redirect failures; the 60-day check catches slower-moving citation decay.
- Compare citation-level data, not just brand mentions. A model can still say your brand name while citing a competitor's page for the actual recommendation — that's a loss even though the brand appears.
An AI visibility audit across major AI platforms run this way turns a migration from a guessing game into a before/after comparison you can put in front of a CMO.
Day 0-7: baseline lock and crawl verification
In the first week, the job isn't measuring AI visibility yet — it's confirming the technical conditions AI crawlers need are intact. Check that GPTBot, PerplexityBot, ClaudeBot and Google-Extended aren't blocked in the new robots.txt, that canonical tags point at the correct new URLs, and that structured data survived the template change.
Most migration-related AI visibility losses trace back to this week, not to content quality. A missed redirect rule that 301s ten legacy URLs to one generic category page is invisible in a manual site check but devastating to a model trying to retrieve the specific page it used to cite.
Day 30: first re-citation check
By day 30, run the frozen prompt set again and compare citation-by-citation against the baseline. This is the point where genuine losses start separating from launch-week noise — a page that briefly 404s during cutover but resolves within days shouldn't show a real drop by now.
If a page that was cited pre-migration is absent at day 30, check three things in order: does the URL still resolve, does it redirect cleanly (one hop, not a chain), and does the migrated version still contain the specific data point or answer the model was citing. Losing a stat table or a pricing breakdown during a content refresh is a common, avoidable cause.
Day 60-90: full share-of-voice comparison
The 60-90 day window is where you measure share of voice against competitors, not just your own before/after. A migration that quietly drops your citation rate while a competitor's stays flat is a real competitive loss, even if your absolute numbers look stable.
This is also the point to check sentiment, not just presence — a model that used to describe your product favorably pre-migration but now pulls from a thin or outdated migrated page can shift tone even while still naming the brand. Monthly AI visibility benchmarking from this point forward keeps the post-migration recovery on a schedule instead of a one-off check.
Why AI visibility drops after a migration
Not every migration causes a drop, and the causes that do show up are consistent enough to check for directly:
- Redirect chains and loops — models following a citation link that hits three redirects before resolving often just drop the reference.
- Schema markup loss — FAQ, HowTo and Product schema frequently don't survive a CMS or template switch and AI engines lean on structured data for extraction.
- URL structure changes without 1:1 redirects — consolidating ten pages into one loses the specific answer a model used to cite.
- Crawler access changes — a new robots.txt or CDN/WAF rule can silently block GPTBot, ClaudeBot or PerplexityBot.
- Content thinning during the rebuild — migrations are a common excuse to "streamline" copy, which strips the exact data points and specifics AI engines quote.
- Domain or subdomain changes — moving from a subdomain to a root domain, or switching TLDs, resets authority signals models weight differently than search engines do.
Baseline your AI visibility before you migrate
Lock citation data now so a post-migration drop is provable, not guesswork.
Does a website migration always hurt AI visibility?
No — a migration only hurts AI visibility when redirects, schema, or crawler access break during the switch, and clean migrations with 1:1 redirects and preserved structured data often show no measurable drop at all. The risk scales with how much the URL structure and content actually change, not with the migration itself.
How long does it take for AI answers to reflect a migration?
AI answers typically start reflecting a migration within 7-30 days for platforms with frequent retrieval like Perplexity, and can take 60-90 days or longer for models that lean more on training-data snapshots. That lag is exactly why a single post-launch check isn't enough — you need the day 7, 30 and 60 checkpoints to see the full timeline.
Should you pause AI visibility tracking during a migration?
No — pausing tracking during a migration is the one mistake that guarantees you can't measure the impact, because you lose the baseline window right when it matters most. Keep the prompt set running through cutover even if the site is unstable for a few days.
FAQ
How do you track AI visibility after a website migration?
Lock a baseline with a fixed prompt set 1-2 weeks before cutover, then rerun the identical prompts at day 7, 30 and 60 post-launch and compare citations, not just brand mentions. This isolates real losses from normal recrawl delay.
What causes AI citations to drop after a migration?
Redirect chains, lost schema markup, blocked crawlers like GPTBot or ClaudeBot, and content thinning during the rebuild are the most common causes in 2026. Domain or TLD changes can also reset authority signals models weight.
How long does it take AI platforms to re-index a migrated site?
Perplexity and other retrieval-heavy platforms often reflect changes within 7-30 days, while models relying more on training snapshots can take 60-90 days. Track on a schedule rather than checking once and assuming stability.
Is a brand mention the same as a citation after migration?
No — a brand mention is the model naming your brand, while a citation links a specific claim to your page. A migration can preserve mentions while losing citations if the cited page's URL or content changed.
Do redirects affect how AI engines cite a page?
Yes — a single clean 301 redirect is usually followed and preserved, but chains of two or more redirects frequently cause AI engines to drop the citation rather than follow it through. Audit redirect chains before and after cutover.
Should AI visibility tracking start before or after a migration?
Before — a baseline captured 1-2 weeks pre-migration is the only way to prove a post-launch drop is caused by the migration and not normal fluctuation. Tracking that starts after launch has nothing to compare against.
Can a migration improve AI visibility instead of hurting it?
Yes — migrations that add structured data, consolidate thin pages into stronger ones, or fix crawler access issues can increase citation rates. The outcome depends on execution, not on whether a migration happened.
One last thing
The pages that recover fastest after a migration are almost never the highest-traffic ones — they're the pages with the most specific, structured answer already sitting on them (a pricing breakdown, a step list, a comparison table). Prioritize checking pages models have stopped citing before you touch the pages that still look fine in your organic dashboard; the citation loss usually shows up weeks before the traffic does.
