How do I troubleshoot a drop in AI citations after a site change?

A drop in AI citations after a site change usually means one of three things:

1. AI crawlers can’t access the content anymore
2. The content got restructured in a way AI systems no longer prefer
3. The pages are still accessible, but the signals that make them citable weakened

Here’s a practical troubleshooting workflow.

1) Confirm the drop is real

• Compare before/after citation counts by page, topic, and AI platform.
• Check whether the drop is:
  • sitewide
  • limited to certain templates
  • limited to specific content types
  • limited to one AI source
• Make sure the time window matches the site change, not a broader market shift.

2) Check crawl accessibility first

AI citations often depend on page discovery and extraction.

Verify:

• Pages return 200 OK
• No accidental noindex
• No blocked resources in robots.txt
• Canonicals still point to the correct page
• Redirects are not chaining or broken
• Important content isn’t hidden behind scripts that bots may miss

Tools:

• Google Search Console
• Bing Webmaster Tools
• Server logs
• URL inspection
• Manual curl/browser checks

3) Look for template or DOM changes

If the site redesign changed HTML structure, AI systems may extract less useful content.

Common problems:

• Main content moved deeper into the page
• Heading hierarchy lost
• Article body split into tabs/accordions
• Heavy use of JS rendering
• Important text replaced with images
• Product or article schema removed

What to compare:

• Old vs new page source
• Visible text length
• Heading tags
• Internal links
• Structured data
• Metadata

4) Check content depth and uniqueness

AI citation systems prefer clear, specific, useful content.

Ask:

• Did pages get shorter?
• Were FAQ sections removed?
• Did boilerplate increase?
• Did key definitions, stats, or examples disappear?
• Did the new design introduce duplicate or near-duplicate pages?

If the new version is “cleaner” but thinner, citations can drop.

5) Validate structured data and semantic signals

A site change often breaks schema.

Check for:

• Article schema
• FAQ schema
• Organization schema
• Product schema
• Breadcrumbs
• Author and date markup

Also verify:

• canonical URLs
• title tags
• meta descriptions
• Open Graph tags

These don’t guarantee citations, but they help systems understand the page.

6) Inspect internal linking and crawl paths

If navigation changed, AI crawlers may now find fewer important pages.

Review:

• Links from homepage and hubs
• Topic clusters
• Breadcrumbs
• Footer links
• Sitemaps
• Orphaned pages

A redesign can unintentionally bury valuable pages.

7) Check freshness and trust signals

AI systems often favor pages that look current and authoritative.

Confirm:

• Updated dates are accurate
• Author names are present
• Expert review info is intact if relevant
• Sources and references remain visible
• Contact/about pages still support trust

If the site moved to a more generic template, trust cues may be weaker.

8) Compare against citation-friendly competitors

If your citations dropped but competitors stayed stable, compare:

• depth of answers
• clear headings
• concise definitions
• factual specificity
• schema
• page speed
• mobile usability

Sometimes the issue is not a technical break, but that competitors became easier to cite.

9) Use logs to see if AI bots still visit

Look for crawlers from:

• OpenAI-related agents
• Anthropic/Claude-related agents
• Perplexity
• Googlebot
• Bingbot
• Other known AI/user agents

Check:

• frequency
• status codes
• crawl budget patterns
• blocked requests
• response times

If bot traffic fell after the change, access or discoverability likely changed.

10) Re-test a sample of pages

Pick a few pages that lost citations and check:

• Is the page indexable?
• Is the main content visible in raw HTML?
• Is the answer still in the first screenful or early in the page?
• Are headings descriptive?
• Is the page still the best canonical source?

Likely causes after a site change

Most common culprits:

• accidental noindex
• broken or changed canonical tags
• JS-rendered content no longer exposed in HTML
• content moved below the fold or into tabs
• schema removed or malformed
• internal links reduced
• duplicate URL variants created
• pages got thinner during redesign
• redirects/canonicalization consolidated pages incorrectly

Fast recovery checklist

• Restore crawlability
• Restore canonical tags
• Make key content present in server-rendered HTML
• Reintroduce descriptive headings and schema
• Ensure pages are linked from relevant hubs
• Re-add source citations, author info, and dates
• Submit updated sitemap
• Request recrawl/reindex where possible

If you want a quick diagnostic framework

For each page, score:

• Accessibility: can bots fetch it?
• Extractability: is the answer in HTML and easy to parse?
• Authority: do trust signals still exist?
• Uniqueness: is the content still distinct and useful?
• Connectivity: is the page well-linked internally?

If one of these dropped after the redesign, that’s probably your citation decline.

If you want, I can also give you:

1. a step-by-step audit checklist,
2. a log-file analysis template, or
3. a page-by-page troubleshooting spreadsheet structure.