Why Your Next Website Migration Needs an AI Visibility Strategy

An SEO website migration is not without risks. Rankings can drop. Tracking can break. Valuable content can go missing, and user journeys can fall apart if redirects aren’t handled properly.

To avoid these issues, most experienced marketing teams involve technical SEO, content, UX, development, accessibility, and analytics specialists before launch.

But one increasingly valuable form of visibility is still being left off too many migration plans.

Customers are increasingly asking AI tools for recommendations, comparisons, explanations and supplier information instead of typing a query into Google.

If AI visibility, recommendations and citations are lost because of a website migration, businesses stand to lose potential customers and leads that come through LLM platforms like ChatGPT, Gemini, and Copilot as well as Google’s own AI Overviews and AI Mode, which now answer a growing share of searches directly on the results page.

That makes AI visibility a launch requirement, not a problem to investigate three months later.

Find out how our GEO and multi-channel services can safeguard your AI visibility before you migrate.

What does AI visibility mean during a website migration?

Referral traffic from AI platforms is an indicator of AI performance but it’s not a reliable measurement of visibility.

AI visibility doesn’t just refer to whether your business appears in AI answers or not; it also concerns how accurately the business is represented and which sources are linked to when it’s mentioned.

These are usually referred to as:

• AI Citations: This is where your company is mentioned in an AI answer. It could be from an informational term and there is no sense that you are being recommended, just cited.
• AI Recommendations: This is when your company is recommended by an LLM for a service or product. This could happen after several clarifying questions from LLMs.
• AI Links: Even when you are cited or recommended, you might not be linked to. Getting a link is important but research shows that even when not linked to, users will go and search manually for your company if recommended.

To protect AI visibility during a website migration, teams need to consider whether LLMs can still:

• Access relevant pages
• Understand what the business does?
• Connect the business to specific services, people, or locations
• Extract a clear fact or answer that can be pulled from the page
• Verify that the source trustworthy
• Corroborate the information against other pages and third-party sources
• Reference the business in relevant answers

AI visibility is an umbrella term for a chain of interconnected signals. During a website migration, any one of these can be disrupted, reducing the likelihood that AI systems will understand, trust, or recommend the business.

How migrations can break AI visibility

AI systems do not rely on one permanent, universally consistent understanding of a business. Their answers may combine existing model knowledge with information they can currently discover, retrieve and corroborate.

Because of this, AI visibility losses are rarely caused by one major technical failure. They come from a series of small migration decisions that change what information remains available, where it sits and how clearly it can be interpreted.

For example:

• Citation-earning pages get removed, redirected incorrectly or absorbed into pages with a different purpose.
• Direct, factual copy gets replaced with creative brand messaging that’s less useful as an AI source.
• Entity relationships that were clear on the old site get separated or weakened by a new architecture.
• Internal linking changes what looks important: Guides, case studies or service pages are moved deeper into the website or become orphaned.
• Structured data gets left out of new templates or stops matching the visible content.
• Third-party references keep linking to pages that no longer exist.

Individually, these look like standard migration housekeeping. Together, they’re how a business that AI systems used to understand well becomes one they don’t.

Some types of migration are more prone to causing AI issues

Not every migration carries the same level of AI visibility risk. The bigger the identity or structural change, the more important protecting AI visibility becomes.

• Domain name changes move every citation, backlink and piece of structured data onto a new address. AI systems don’t just need a redirect map; they need to re-learn that the new domain is the same business, with the same authority, as the one they already trusted.
• Rebranding changes what AI systems are trying to match in the first place. If the business name, positioning or messaging shifts while old citations, directories and press coverage still reference the previous brand, AI answers can end up blending old and new identities.
• Large-scale site restructures touch URL hierarchy, internal linking and content grouping across the whole site at once, rather than page by page. This multiplies the risk of orphaned content and broken entity relationships across dozens of citation-earning pages

The AI visibility migration checklist

The best time to protect AI visibility is before URLs, templates and content begin to change. Any AI visibility migration plan should sit alongside the existing SEO, content, UX, development and analytics workstreams.

The following checklist covers what to benchmark before the migration, what to preserve or improve during the build and what to test and monitor after launch.

1. Benchmark current AI visibility

Before you touch the website, record where the business currently stands across priority prompts covering brand information, core services, commercial comparisons and common customer questions.

This phase is about identifying a visibility benchmark not producing a definitive ‘AI ranking’. Without the record, you have no way to tell how successful your efforts to protect AI visibility were.

Build a fixed list of prompts covering priority categories, then run each one manually across ChatGPT, Gemini, Copilot, Perplexity, Claude and Google’s AI Overviews and AI Mode.

Alternatively, tools like SEMrush now offer AI visibility tracking that can run this benchmark for you, rather than testing every prompt by hand.

H3: 2. Identify citation-earning pages and facts

Find the pages, statistics, definitions, case studies and company facts that already show up in AI answers or make strong citation candidates.

Treat them the way you’d treat your best-performing organic landing pages in a migration and decide whether each will be retained, updated, consolidated, redirected, replaced or removed.

You’ll usually find these pages in three places:

• URLs cited in your benchmarking prompts
• AI referral traffic already showing up in Google Analytics or Search Consol
• Server-log activity from recognised AI crawlers such as GPTBot, Google-Extended or PerplexityBot.

Cross-reference your list against pages with strong organic traffic or backlinks too, since a page can be citation-worthy before an AI system has cited it.

3. Map important brand entities

List the entities AI systems need to connect accurately, from the organisation itself to its services, products, locations, founders and accreditations.

Check that those relationships stay explicit in the new site’s copy, navigation and structured data so that LLMs can understand the connections. Don’t rely on a reader – human or machine – to infer a relationship from vague brand language.

Build this out as a simple entity map or spreadsheet: one row per entity, with the pages, schema and internal links that currently establish the connection.

Check it against your existing Organisation and Person schema, your Google Business Profile or Knowledge Panel if you have one, and any Wikipedia or Wikidata entries, since these are common reference points AI systems corroborate against.

4. Review URL and redirect changes

Redirects preserve access, but they don’t guarantee the replacement page contains the information that made the original one useful. Map every important old URL to the most relevant new destination to make sure routes to the right information are intact.

Start by crawling the existing site with a tool like Screaming Frog to export a full URL list, then cross-reference it against your backlink profile (available through SEMrush or Ahrefs), your analytics data and the citation list from steps one and two.

Pay particular attention to URLs with: 

• Strong backlinks
• Press coverage
• Existing AI citations
• Research or statistical content
• Product specifications
• Definitions
• Case studies
• Evergreen guides

5. Preserve headings, context and answerable passages

Clear headings help users and AI systems understand what each section covers. They should be by direct, self-contained answers that don’t force a reader to piece information together across the page.

You don’t need to rewrite every section for an AI system, just keep useful information clear and unambiguous.

Run a side-by-side content audit of old versus new page templates, checking each H1, H2 and H3 still exists in some form and hasn’t been replaced with vague marketing copy.

For FAQs and definitions specifically, check that the answer appears in the first sentence or two after the question, rather than being buried under supporting paragraphs, and mark up FAQ content with FAQPage schema.

6. Retain or improve structured data

Take stock of what’s on the existing website before templates get replaced. Extract the schema currently live on the site and document what’s on each template.

Pay particular attention to:

• Organisation
• LocalBusiness
• Person
• Product
• Service
• Article
• BreadcrumbList
• Event
• JobPosting
• Review
• VideoObject
• sameAs

Once the new site is built, validate every key page with Google’s Rich Results Test and the Schema Markup Validator, checking specifically that entity names and IDs match what you mapped in step three, and that nothing still points to a URL the migration has retired.

7. Test crawler access and rendering

Review your robots.txt, CDN settings, security tools, bot management rules and server responses.

Confirm that the AI crawlers you want visiting can reach core pages, without hitting logins, noindex directives carried over from staging, or firewalls blocking them unintentionally.

To do this, check robots.txt line by line for each named AI user-agent you care about, such as GPTBot, Google-Extended, PerplexityBot and ClaudeBot, since these are often controlled separately from general search crawlers.

Test actual access by fetching key pages with those user-agent strings (using a tool like Screaming Frog’s custom user-agent setting, or a simple curl request) rather than assuming the robots.txt file is comprehensive, then check server logs after launch to confirm those crawlers are actually visiting, not just permitted to.

AI search and model training are not always governed by the same crawler or permission.

A business might choose to allow its content to be retrieved for current AI search answers while preventing it from being used for model training. Confirm the business’ policy and configure crawler permissions accordingly, rather than treating all AI bots as one category.

8. Reduce reliance on JavaScript for critical information

Ask your developers to check what’s available before client-side JavaScript executes.

Interactive functionality can still lean on JavaScript, but if the facts themselves only load after an interaction, access becomes far less reliable across different crawlers and retrieval systems, not just Google’s.

Prioritise server rendering or static HTML for: 

• Company descriptions
• Service information
• Product details
• Key facts
• Pricing information
• Locations
• Contact details
• FAQs
• Author information
• Internal links

Check this by viewing the page source (not just the rendered page) or disabling JavaScript in your browser and reloading key pages. If the critical facts vanish, so will an AI crawler’s ability to read them.

Where JavaScript is unavoidable, ask developers to implement server-side rendering or static generation specifically for the sections carrying core facts, rather than rendering the whole page client-side by default.

9. Audit third-party references

Identify the authoritative external pages that mention or link to your business. This can include press coverage, trade publications, partner and supplier sites, directories, review platforms, research reports, event pages and social profiles.

Pull this list together with a backlink audit in Ahrefs or SEMrush, filtered for the URLs due to change. Also, conduct a manual search for your brand name to catch mentions that don’t carry a live link.

Where you can’t update digital PR references directly, make sure the old URL redirects to a closely matched page to help LLMs corroborate information. For sources you have a relationship with, such as partners, press contacts or directories you’re listed on, reach out directly and ask them to update the link.

10. Create a post-launch AI monitoring plan

AI visibility should sit alongside ranking, organic traffic, conversions and crawl errors in your post-launch reporting, not off in its own silo.

To monitor AI visibility after launch, track:

• Changes in brand mentions across agreed prompts
• Changes in cited sources
• Accuracy of company descriptions
• AI referral traffic
• Landing pages receiving AI referrals
• Server-log activity from recognised AI crawlers
• Indexation and rendering issues
• Broken external references
• Competitor visibility
• New or lost citations

Put the same benchmarking prompts from step one on a recurring schedule, whether that’s manual re-testing or an AI visibility tracking tool, and add AI crawler activity to your regular server-log review.

Schedule one-week, one-month and three-month check-ins alongside typical post-migration SEO and analytics reporting. AI results can vary between users, models and even repeated prompts on the same day, so look for patterns rather than treating any single answer as a fixed ranking.

Our AI Visibility Migration Checklist covers all ten in full, with the specific checks, tools and questions to run through at each stage.

Download the full AI Visibility Migration Checklist

AI visibility cannot be owned by one team

AI visibility touches almost every team involved in a website migration, which is exactly why it can fall through the cracks.

• SEO teams understand crawlability, redirects and information architecture.
• Content teams understand how facts, expertise and evidence are communicated.
• Developers control rendering, templates and whether a crawler can reach anything at all.
• PR and digital PR teams influence the third-party sources AI systems lean on to corroborate a claim.
• Analytics teams work out whether that visibility is actually turning into referral traffic and leads.
• Accessibility teams help ensure important information remains available through clear structure, semantic HTML and accessible interfaces.
• And brand and legal teams decide what can be claimed, and how the business gets represented.
• No single person owns all of that. Which means AI visibility needs a seat on the main migration spec from day one, not a line item on the SEO team’s to-do list, tackled after the fact.

What should happen before migration work begins?

Before URLs, templates and content are finalised, a pre-migration AI visibility audit should be conducted to answer the following questions:

• What the business is currently known for?
• Which pages and sources support that understanding?
• Which information absolutely has to survive the move?
• Which technical barriers need resolving before launch?
• How success will be measured afterwards?

A successful migration does not simply move pages. It preserves the relationships, evidence and authority that help both people and AI systems understand why the business deserves to be found.

Planning a website migration?

Edge45 will help you protect organic performance, brand authority and AI visibility throughout a website migration, so you launch with confidence instead of finding out three months later what you lost.

Speak to our team today.