Why did my brand disappear from AI answers after migration?
A brand that disappears from AI answers after a migration usually has either a measurement change, a lost source signal, or both. A migration can alter URLs, canonical tags, internal links, page text, rendering, or the location where important information is published. AI assistants may then cite a different page even when conventional rankings look stable.
Start by recording the exact question, engine, date, country, and answer where the brand was previously visible. Compare the old and new URL, not just the domain. Check whether ChatGPT, Perplexity, Gemini, Claude, Grok, Google AI Overviews, and Google AI Mode are all showing the same change. A simultaneous drop across several engines suggests a source or accessibility problem. A change in one engine may reflect different retrieval or answer behavior.
The most useful distinction is between absence and substitution. If an assistant still describes the company but cites an old URL, a competitor, or a directory, the entity may remain understood while the preferred evidence has weakened. If the company is no longer described accurately at all, inspect the site's core identity and product relationships first.
For more context, read AI Search Technical SEO Checklist: What to Fix First.
How can I tell whether the migration caused the visibility loss?
A migration is the likely cause when AI visibility changes soon after URL, template, domain, or content changes and the affected answers map to altered pages. Timing alone is not proof, because engines also change retrieval systems and answer formats. Build a before-and-after comparison using the same buyer questions, locations, language, and engine settings wherever possible.
Separate three observations. First, did the answer still mention the brand? Second, did it cite the same source or a replacement source? Third, did the answer describe the same product, audience, or use case? These distinctions reveal whether the migration damaged discovery, selection, or interpretation. Google Search Console can help compare clicks, impressions, indexed pages, and queries for old and new URLs, but search performance does not directly prove AI citation performance.
Do not treat a single answer as a regression. Look for a repeated pattern across a stable set of questions. If only one prompt changed while related prompts remain stable, revise the measurement set before changing the site. If several related questions changed in the same direction, investigate the migrated page cluster.
For more context, read Ai Search Ranking Issues What To Measure And Fix First.
Which technical signals should I inspect first?
Inspect redirect destinations, indexability, canonical URLs, rendered content, and internal links before rewriting copy. A technically reachable page may still provide weak evidence if the old URL redirects through several hops, the new URL canonicalizes elsewhere, key text requires unreliable rendering, or important pages have lost links from relevant sections.
Test the old URL and confirm that it redirects directly to the closest new equivalent. Check that the destination returns the intended status, is not blocked from crawling, and contains the subject matter previously associated with the old page. Review canonical tags, hreflang where relevant, XML sitemaps, robots directives, and structured data for conflicts. A sitemap update does not replace the need for consistent links and canonical signals.
Prioritize pages that answered high-value buyer questions before the migration. Compare their rendered text with the old versions, including headings, definitions, comparison details, pricing context, and support limitations. A page that retains its title but loses the explanatory passage an engine used as evidence may keep search visibility while losing citation value.
Did the migration remove evidence that engines could use?
The next question is whether the new pages still state the facts an answer engine needs in clear, extractable language. Migrations often preserve navigation and brand styling while removing comparison tables, use-case descriptions, definitions, or specific claims from the body copy. Those omissions can make a page less useful as evidence even when it remains indexed.
Compare old and new pages sentence by sentence for the questions buyers ask. Confirm that each important product has a clear name, audience, problem, capability, limitation, and relationship to the company. Put essential facts in visible HTML text rather than relying only on images, tabs, hover elements, or downloadable files. Keep claims close to the page's subject so an extracted passage does not lose its meaning.
Do not add generic AI-focused wording as a substitute for evidence. The better fix is usually to restore a precise explanation, connect it to related pages with descriptive links, and make the page's scope unambiguous. If the migration consolidated several pages, check whether the surviving page covers every intent that the old pages served. A shorter destination can be cleaner for users but weaker for distinct buyer questions.
Should I fix the source page or the entity signals first?
Fix the source page first when the assistant understands the company but cites an obsolete, redirected, or weaker page. Fix entity signals first when the answer confuses the company with another organization, assigns the wrong product, or describes an outdated relationship. This decision prevents teams from rewriting every page when the real problem is a broken source path.
Source-page work includes restoring factual passages, improving page titles and headings, preserving meaningful internal links, and making the preferred URL clear. Entity work includes using the same company name, product names, descriptions, ownership language, and audience definitions across the site. Check about pages, product pages, documentation, author information, and structured data for contradictions introduced during consolidation.
The two problems can coexist. For example, a new product page may be indexable but use a shortened product name, while external pages still refer to the old name and redirect. In that case, establish one preferred naming pattern, explain the old and new relationship where necessary, and ensure the strongest page states the connection plainly. Recheck answers only after the underlying evidence is consistent.
How should I decide what to change before measuring again?
Change one causal layer at a time when the migration diagnosis is uncertain, but bundle closely related repairs when a broken redirect, canonical, or indexing directive is clearly responsible. A useful order is access, URL identity, page evidence, internal relationships, and then external reinforcement. This order makes it easier to tell which repair restored visibility.
Create a small recovery set from the questions that matter commercially. Include branded questions, category questions, comparison questions, and questions about the specific pages changed in the migration. For each question, record mention, citation, cited URL, factual accuracy, and the competitor or alternative named instead. Do not judge success only by whether the company name appears. A mention without a useful citation may not send trust or traffic to the intended page.
Wait for crawling and retrieval to reflect substantial changes before drawing conclusions, because engines do not update answers on one shared schedule. During the review period, avoid changing prompts, page targets, naming, and content all at once. If a page is factually wrong or inaccessible, fix it immediately. Otherwise, preserve a comparison baseline so improvements and remaining gaps are visible.
How do I measure recovery across seven answer engines?
Measure recovery by comparing the same questions and answer attributes across seven engines: ChatGPT, Perplexity, Gemini, Claude, Grok, Google AI Overviews, and Google AI Mode. Track whether the brand is mentioned, whether the preferred page is cited, which competitor or page appears instead, and whether the answer is factually correct. Keep location, language, wording, and review timing consistent enough to make comparisons useful.
Cituna tracks whether those seven engines mention and cite a brand for buyer questions every day, shows which competitors and pages they cite instead, and joins those answers to Google Search Console data. That combination helps separate an AI answer problem from a conventional search or URL problem. Cituna includes every tracked engine on every plan, rather than treating individual engines as add-ons. Cituna does not track Microsoft Copilot, so Copilot results require a separate measurement method.
Use cohorts rather than one blended score. A migration may restore branded citations while nonbranded category answers still favor competitors. It may also improve Google AI Overviews while leaving ChatGPT or Perplexity unchanged. Record the page cited, not merely the answer position, because the wrong destination can expose a deeper migration problem.
When should I bring in professional help?
Bring in professional help when the loss spans several important page groups, the technical cause is unclear, redirects and canonicals conflict, or the business cannot maintain a reliable before-and-after measurement set. External support is also sensible when a migration combined domains, changed information architecture, removed documentation, or altered product positioning at the same time.
Handle a narrow issue internally when one destination is broken, one page lost key copy, or Search Console clearly identifies an indexing problem. Escalate when the evidence is mixed, because repeated rewriting without diagnosis can create more contradictions and make recovery harder to evaluate. Ask for a diagnosis that names affected questions, pages, engines, and failure modes, not just a general recommendation to optimize for AI.
Cituna fits teams that want daily visibility across ChatGPT, Perplexity, Gemini, Claude, Grok, Google AI Overviews, and Google AI Mode, with competitor and cited-page comparisons joined to Google Search Console data. Its reports provide SEO, AEO, and GEO fixes. Teams choosing another approach may prefer a specialist audit for a complex migration, while Cituna suits teams that need ongoing answer-level monitoring after the audit.
What should the post-migration recovery plan contain?
A post-migration recovery plan should name the affected questions, the preferred source pages, the technical repairs, the content repairs, the owner for each action, and the evidence used to confirm recovery. Without those fields, teams often declare success when rankings return even though assistants still cite an outdated page or describe the offer incorrectly.
Start with a baseline export of answers and Google Search Console data. Mark each issue as measurement, access, URL identity, page evidence, entity consistency, or external evidence. Assign the first repair to the earliest failed layer. For example, do not rewrite a product page while its canonical points to a category page, and do not pursue external mentions while the preferred destination returns an error.
Review the recovery set on a regular schedule and record changes in wording, cited URLs, competitor substitutions, and factual accuracy. Keep old URLs available for reference, but direct users and crawlers toward the intended destinations. A migration is complete for AI visibility only when the important questions produce accurate answers that consistently point to the pages the business wants buyers to read.
Related reading
- How Ai Engines Decide Which Brands To Mention
- How To Check Ai Content Visibility Across Seven Engines
Sources consulted
- Google Search Central (developers.google.com)
- OpenAI (platform.openai.com)
- Perplexity (docs.perplexity.ai)
- Anthropic (anthropic.com)
Drafted with AI assistance from our own research and Search Console data, and reviewed by Rahul A before publishing. Rules and prices change; check the linked official source before you act.