Site migrations without losing AI citations
By Abhijay Tondak, Founder & CEO · Updated June 30, 2026 · 7 min read
To migrate a site without losing AI citations, preserve the things engines relied on to cite you: keep URLs stable or 301-redirect them one-to-one, keep the cited content and its answer-first structure intact, and make sure the new site is fully crawlable so engines can re-index it. Citations break in a migration for the same reasons rankings do — broken redirects, changed content, or blocked crawling — so the playbook is disciplined preservation plus a thorough post-launch re-crawl check.
Key takeaways
- Citations depend on stable, retrievable URLs — map every old URL to its new home with a 301.
- Preserve the cited passage: if you rewrite the answer during migration, you can lose the citation.
- Don't block the new site in robots.txt or ship a noindex by accident — the classic migration killers.
- Re-submit your sitemap and confirm AI crawlers can reach the new pages after launch.
- Expect a re-indexing lag; monitor citations through it rather than panicking on day one.
Why migrations put citations at risk
When you change your CMS, domain, or URL structure, you're altering the exact things AI engines used to find and trust your content. If a cited URL now 404s, the citation has nowhere to point. If the content moved but the answer-first passage got rewritten, the thing the engine quoted may no longer exist. And if the new site accidentally blocks crawlers, engines can't re-index it at all. Migrations are high-risk precisely because they touch retrieval, content, and crawlability at once.
Map and redirect every URL
The backbone of a safe migration is a complete URL map: every old URL paired with its new destination, served via a permanent 301 redirect. Don't bulk-redirect everything to the homepage — that destroys the equity and the citation target. Redirect each cited page to its true equivalent so the authority and the reference transfer cleanly.
- Inventory every indexable URL before you start (sitemap + crawl).
- 301 each old URL to its closest new equivalent — one-to-one, not many-to-home.
- Test the redirects post-launch; a redirect chain or loop is as bad as a 404.
- Keep redirects in place long-term — engines and links rely on them for months.
Put this into practice
See how your site performs across AI engines with a free visibility audit — takes 2 minutes, no credit card.
Run your free auditPreserve the cited content, not just the page
A migration is tempting as a 'while we're here, let's rewrite everything' moment — but for your cited pages, that's where citations die. The specific answer-first passage an engine quoted is an asset; preserve it. If you must improve a cited page, change it deliberately and treat it as an experiment, not a casual rewrite buried in a 500-page migration. Keep the structure (headings, FAQs, schema) that made it extractable.
Don't break crawlability
The most common self-inflicted migration disaster is shipping the new site with a staging robots.txt that disallows everything, or a leftover noindex tag. Either one tells engines to ignore the whole site, and citations evaporate as pages drop out of the index. Before and right after launch, verify robots.txt allows crawling (including AI crawlers), confirm no stray noindex, and check that structured data survived the move.
Re-index and monitor through the lag
After launch, actively help engines re-discover the new site: re-submit your sitemap, confirm AI bots are crawling the new URLs (watch your server logs), and spot-check that cited pages resolve and render. Then be patient — re-indexing and answer regeneration take time, so a temporary dip is normal. Monitor your baseline citation set through the transition; a sustained drop after the lag means a redirect or crawl issue to hunt down, not a reason to panic on day one.
Frequently asked questions
Will I lose citations during a migration no matter what?
A short, temporary dip during re-indexing is normal even on a clean migration. Permanent loss is avoidable — it comes from broken redirects, rewritten cited content, or blocked crawling, all of which the preservation playbook prevents.
Should I redirect old URLs to the homepage if there's no exact match?
No — redirect to the closest relevant page. Mass-redirecting to the homepage drops the citation target and wastes the authority. Only pages with truly no equivalent should 301 to a sensible parent or category.
How long should I keep migration redirects in place?
Indefinitely, or at minimum a year or more. Engines, links, and citations rely on them well after launch; removing redirects early reintroduces the 404 problem you migrated to avoid.
What's the number-one migration mistake for AI search?
Shipping with crawling blocked — a staging robots.txt disallow or a leftover noindex. It silently removes the whole site from the index, taking every citation with it. Verify crawlability before and immediately after launch.
Put this into practice — free.
Get your free AI-visibility audit and see where engines find you today.
More from this topic
Keep building your expertise with related GEO content in the same cluster.
AI SEO tools comparison: evaluation criteria for 2026
Compare AI SEO tools by what matters: citation tracking, content generation, structured data audit, and measurement. Here's how the categories stack up and what to prioritize.
ReadGEO SEO tool: what to look for and how to choose
A GEO SEO tool combines traditional search optimization with AI citation tracking. Here's what capabilities matter, how the category works, and how to evaluate one.
ReadGEO platform comparison 2026: how the tools stack up
A 2026 comparison of GEO platforms: how citation tracking, content generation, and technical audit capabilities differ — and which combination fits your team.
Readhreflang and international GEO
hreflang tells engines which language/region version of a page to serve. Here's how it works, why it matters for international AI search, and how to implement it without the common mistakes.
ReadJavaScript rendering and AI crawlers
Many AI crawlers don't execute JavaScript, so content rendered client-side can be invisible to them. Here's how JS rendering affects AI citations and how to make content crawlable.
ReadMaking single-page apps (SPAs) crawlable for GEO
Client-rendered SPAs are notoriously hard for AI crawlers to read. Here's how to make a single-page app citable - SSR/prerendering, real URLs, and content in the initial HTML.
ReadKeep reading
Is your site agent-ready?
Most sites score under 30. Check yours in seconds — get a 0–100 agent-readiness score and a prioritized fix list.