SEO Playbook That Killed AI Citations (Case Study)
Last updated: July 2, 2026
Maria Dykstra is an AI Visibility Architect who has diagnosed algorithmic authority failures for 50+ B2B companies.
She built global ad systems at Microsoft that drove $2B in revenue across 1B+ ads per month. She ran TreDigital for 13 years across Fortune 500s and startups. She is embedded with agentic AI companies to translate their infrastructure into go-to-market strategy.
← Guide 4: Why Does AI Ignore My Content? Fix Citation Invisibility
A year of AEO work built a regulated-industry client to roughly 860 cited URLs and 38,000 Bing AI citations across six months. A content consolidation phase using standard best practice cut citation volume by more than half. The shift happened in the retrieval environment, not in the playbook. The data caught it. The recovery sequence and the framework that emerged are below.
- ~860 cited URLs and ~38K citations earned over 12 months in a regulated services category
- A standard consolidation phase cut citation volume by ~55% within 30 days
- The shift was in the AI retrieval environment, not in the AEO playbook
- A URL earning AI citations is a Citation Asset. Treat it as a retrieval asset, not supporting content.
- Recovery: identify the Citation Assets, restore them at original URLs, keep pillars alongside
- Part 2 publishes August 2026 with 60-day recovery data from this engagement
By Maria Dykstra · AI Visibility Architect · Published June 23, 2026
The work that worked
This case study comes out of a long-term engagement in a regulated services category. A vertical where trust signals carry heavy weight and AI systems are conservative about which sources they cite.
Over the prior 12 months, the work had been compounding. We took the site from minimal AI citation footprint to roughly 860 distinct URLs being cited across the most recent six-month window, totaling close to 38,000 Bing AI citations. The trajectory was up and to the right across the standard metrics.
The playbook was AEO-led from the start. Schema implementation. Answer-first content structure. FAQPage markup across high-intent pages. Inverted pyramid drafting. Semantic anchor consistency on the proprietary terminology. The standard AEO playbook applied with discipline, against the standard AEO measurement cadence.
It worked. Until one phase of the work didn’t.
The consolidation phase
In late March 2026, we ran a content consolidation pass. The site had 37 supporting blog posts across a few topic clusters. Each post answered a specific question. The consolidation strategy was to roll them into a smaller set of pillar pages and 301 the supporting posts forward.
This is textbook topic cluster strategy. Hub-and-spoke architecture. Internal link equity concentration. Pillar pages as topical authority anchors. It is the dominant content strategy advice across SEO and AEO best-practice writing from 2018 forward.
The redirects deployed across March and April. Each one was set up cleanly. Each pillar page was well-written. The supporting posts stayed published in WordPress, intercepted by 301s before any visitor or crawler could reach them directly.
Within 60 days, the citation metric the site had been climbing for a year reversed.
What we caught in the data
Bing AI Performance was already part of the standard measurement cadence. Weekly review, monthly trend analysis, quarterly cited-pages audit. The monitoring caught the drop early.
The before-and-after, sized to compare cleanly.
| Window | Approx. citations | Cited pages |
|---|---|---|
| Peak pre-consolidation (March 2026) | ~7,500/month | ~250 |
| Current 30 days (late May to late June) | 3,400 | 127 |
| Change | ~55% drop | ~50% drop |
The pattern was specific. Nine of the top 20 most-cited URLs over the prior six months had disappeared from the 30-day list entirely. The pages had not been deleted. They were just no longer being served as themselves.
How we analyzed the drop
The analytical process took longer than the recovery did. The data tells a story when you stop interpreting it through the playbook that produced it.
We pulled the full Bing Cited Pages export for the prior six months. 859 URLs with citation counts. We pulled the 30-day export. 127 URLs. The set difference identified every URL that had dropped out of citation between the two windows.
For each dropped URL, we ran the same diagnostic. Open the URL in an incognito tab. Note the HTTP status code. Note any redirect destination. Compare the destination’s content against the original URL’s topic. Flag whether the redirect target answered the same question or a different one.
The pattern that emerged was specific to the consolidation phase. URLs that 301’d to a pillar covering broader content lost citation weight. URLs that 301’d to a pillar covering closely matched content held some citation weight or were beginning to recover.
The expansion was the next step. Nine URLs in the original top-20 audit had dropped. But the consolidation phase touched more than nine URLs. We expanded the audit across every redirect that deployed in the March-April window. The full impact set was 37 URLs across five topic clusters.
Each one followed the same pattern. Live in WordPress. Intercepted by a 301. The original answer asset still existed at the original URL, just unreachable.
Why standard consolidation breaks AI citations
This is the part where I have to be careful about what I can prove versus what I can infer.
What I can prove: 37 URLs were redirected in late March. Bing AI citations on those URLs and across the broader cluster dropped sharply within the same window. The decline is too tightly timed to be coincidence.
What I can infer from the data: AI citation systems appear to treat the URL as more than a disposable delivery path. They seem to associate citation eligibility with the URL and the specific extractable answer that lived at it. A redirect to broader content appears to break that association.
The mechanism likely runs something like this. Bing’s AI extractor finds candidate URLs through retrieval, then extracts a clean answer from each, then weights them for citation. When a URL stops serving the answer the extractor learned to cite, the citation drops. The new pillar page has to earn extraction confidence from scratch against a different set of queries that probably do not match cleanly to what the supporting post originally answered.
This is Citation Invisibility. Layer 4 of the Algorithmic Authority Stack. The failure mode where AI can find your content but cannot extract a clean answer worth citing. The 37 URLs were not invisible to Bing’s crawler. They were invisible to Bing’s extractor, which appears to be a different system with different rules.
The retrieval shift that outpaced the playbook
The standard AEO playbook assumed a unified retrieval environment. Improve answer extraction quality and topic-cluster authority together. Consolidate supporting content into pillar pages to concentrate topical authority. Watch both Google and AI surface rewards compound.
That assumption held when the playbook was written. It is holding less reliably now. AI citation extractors appear to be optimizing for specific answer extraction at the URL level. Topic-cluster authority still matters at the broader category level. The two signals are reinforcing on some metrics and divergent on others.
This is what the Great Decoupling looks like at the URL level. Human visibility and machine visibility used to be the same system. They are not anymore. The retrieval shift is happening faster than playbook documentation can catch up to it.
The signal that we needed an additional step in the consolidation playbook was not visible in the standard AEO best-practice writing in March. It is visible now, in this data.
The Citation Asset framework
The category we needed and did not have was the Citation Asset.
Citation Asset. A URL currently earning AI citations from any major retrieval system: Bing AI Performance, Perplexity, ChatGPT, or Google AI Overviews. Once a URL earns citation weight, it stops being interchangeable supporting content. It becomes a retrieval-system asset producing compounding visibility. Citation weight does not transfer to the new URL by default when a Citation Asset is redirected or restructured.
The framework adds one step to the standard consolidation playbook. Before any redirect deploys, segment the candidate URLs into two groups.
Group A: URLs earning meaningful AI citations. These are Citation Assets. Leave them at their original URLs serving their original content.
Group B: URLs not earning AI citations. Standard consolidation candidates. Redirect into pillar pages as planned.
The pillar pages get built alongside the Group A URLs, not on top of them. Both retrieval systems get served. Neither loses visibility.
Do not treat a cited URL as supporting content. Treat it as a retrieval asset.
How we recovered
The recovery was operationally simpler than the diagnosis. Because the original posts were still published in WordPress, the fix was not restoration. It was clearance.
The sequence ran in this order.
First, we verified each affected post existed at its original URL with its original content. WordPress admin, post-by-post check, status confirmed Published, slug matched, content intact.
Second, we documented every redirect rule pointing away from those URLs. Source URL, destination URL, redirect type, plugin rule ID. The documentation step matters because some redirects had cascade dependencies we needed to understand before deleting anything.
Third, we deleted the redirects in batches. After each deletion, we tested the source URL in an incognito window to confirm it returned 200 with the original content rather than 301-ing to the pillar.
Fourth, we resubmitted the XML sitemap to Bing Webmaster Tools and Google Search Console. Bing accepts sitemap signals quickly. Google sometimes needs a manual nudge, so we used URL Inspection to request indexing on the highest-citation URLs.
Fifth, we set the monitoring cadence for recovery. Weekly Bing AI Performance check on the 30-day rolling window. Specific URL tracking for the highest-value Citation Assets. Recovery targets at 30, 60, and 90 days.
Total execution time across all 37 URLs: under two hours. The pillar pages stayed live at their own URLs. Both surfaces now exist on the site.
What we’re doing differently now
The Citation Asset audit is now part of the standard consolidation playbook on every engagement. Before any redirect deploys, the candidate URLs run through the Bing AI Performance Cited Pages check and the Perplexity citation check on top buyer queries. URLs earning citations stay live.
The monitoring cadence is also updated. Cited pages count is now a leading indicator we track weekly, not monthly. A sustained drop in cited URL count is a signal to investigate before the citation count drops with it. In this engagement, the cliff in cited pages was visible roughly two weeks before total citations fully bottomed out. That window matters for catching the next retrieval shift earlier.
In Wave 1 of the Algorithmic Authority Index, 19 of 20 companies showed at least one Citation Invisibility failure in their existing content. Most were caused by consolidation, rebrand, or migration projects executed before AI citation systems became this central to retrieval. The patterns repeat across industries. The recovery sequence is roughly the same: identify the Citation Assets, restore them at their original URLs, accept that the broader content strategy may need to coexist with the specific content rather than replace it.
What comes next
Part 2 publishes in August 2026 with 60 days of recovery data from this engagement. Three questions get answered with real numbers.
- How many of the 37 URLs recover their citation volume, and how fast.
- Whether the pillar pages continue ranking in Google after the supporting posts return.
- Whether Domain Rating recovers in parallel or lags the citation recovery.
The clean version of the story is the one where everything recovers and the new framework is validated. The honest version may be messier. We will publish whatever the data shows.
Diagnostic · 48 Hours
Is your site losing AI citations to a redirect or consolidation project?
The AI Visibility Snapshot tests Layers 1, 3, and 6 across ChatGPT, Perplexity, and Gemini in 48 hours. Identifies which Citation Assets you have, which ones are at risk, and which failure modes are active in your specific company.
Run the Snapshot →FAQ
How can I tell if a URL is a Citation Asset before redirecting it?
Open Bing Webmaster Tools. Navigate to AI Performance. Set the date range to 6 months. Review the Cited Pages list. Any URL on that list is a Citation Asset.
Cross-reference with Perplexity by running your top 10 buyer queries and noting which of your URLs appear in citations. If a URL appears in either source, do not redirect it during a consolidation project without restoring it later or accepting the citation loss as a deliberate trade.
What’s the difference between Google’s view of a 301 redirect and Bing AI’s view?
Google treats a 301 as authority transfer. Most ranking equity moves to the new URL. Bing’s AI citation system appears to treat URL changes differently.
When a previously cited URL stops serving its original content, citation weight does not migrate cleanly to the new target. The new URL has to earn citation eligibility against a different set of queries. This is observed behavior, not documented Bing architecture. The pattern is consistent across the case in this post and similar engagements.
If the AEO playbook was sound, how did it produce this outcome?
The playbook was sound for the retrieval environment it was built within. Schema implementation, answer-first content, semantic anchor consistency: all still correct.
The shift happened in how AI extractors treat URL-level answer eligibility versus topic-cluster authority. The consolidation move that strengthens topic-cluster authority appears to weaken URL-level answer eligibility. The playbook needed an additional Citation Asset audit step. That step did not exist when the consolidation was planned.
Can I recover AI citations after a consolidation has already happened?
Often yes. If the original supporting posts still exist in WordPress, deleting the redirects and restoring the URLs to serve their original content is the fastest path.
The pillar pages can stay live at their own URLs. Both surfaces coexist. Recovery timelines: 2 to 4 weeks for first re-cited URLs in Bing, 8 to 12 weeks for fuller recovery, longer for Domain Rating to update on third-party tools.
How do you catch a retrieval shift before it costs you citations?
Continuous monitoring of Bing AI Performance, Perplexity citation tracking on top buyer queries, and Google Search Console coverage trends. Weekly review of cited pages lists.
Any sustained drop in cited URL count is a signal worth investigating before the citation count drops with it. The cliff in this engagement was visible in Bing’s cited pages metric two weeks before the citation count fully bottomed out. Without the monitoring cadence in place, the lag between cause and detection extends, and recovery starts later.
Related Reading
- Why Does AI Ignore My Content? Fix Citation Invisibility. The Layer 4 guide covering the failure mode this case study demonstrates at scale.
- What Is the Difference Between SEO, AEO, and GEO?. The retrieval-system framework underneath why one playbook no longer covers all three.
- Non-Commodity Content and AI Visibility. Why generic pillar pages get cited less often than specific supporting content.
- The Algorithmic Authority Stack: 7 Layers Between You and AI Visibility. The full diagnostic framework.
- My Rebrand Was Six Months Ago. AI Still Uses the Old Name.. The same parametric-versus-live-retrieval problem at the entity level instead of the URL level.
Is your company invisible to AI?
Six questions, about 90 seconds. Find out which of the seven layers is breaking first, and whether you are failing to be retrieved or failing to be cited.
Take the AI Visibility Test
