The AI Visibility Gap: An Original Data Study by Digital Authority Partners
What You’ll Learn in This AI Visibility Study
Most businesses assume that earning a citation or brand mention in an AI answer is the goal of AI visibility strategy. Our data shows the opposite. Getting cited is the easy part. Staying cited is the real challenge.
Digital Authority Partners tracked 1,127 unique URLs across ChatGPT, Perplexity, Gemini, Copilot, and Google AI Overviews over three collection waves spanning six weeks. We ran the same 30 queries, covering five intent types and six industries, and recorded every citation, brand mention, and source URL at each wave. The result is one of the first controlled, longitudinal studies of AI citation behavior across multiple platforms.
What we found challenges nearly everything the industry currently believes about AI visibility. The sources AI engines cite today are not the sources they cite two weeks from now. Our study shows that 66% of the URLs AI engines cited in week 1 were gone by week 4.
The majority of AI-cited content does not rank anywhere in the top 20 results on Google or Bing. Each AI platform draws from its own universe of sources with almost no overlap between them. And when AI engines recommend a brand by name, 89% of the supporting citations point to third-party media rather than the brand’s own website.
This study introduces the concept of Citation Retention Rate (CRR), a metric for measuring whether AI citations persist or churn over time, and documents a phenomenon we call “dark matter,” AI-cited content that is invisible to every traditional SEO monitoring tool. Inside the full report, you will find the complete platform-by-platform data, vertical-specific recommendations, and a framework for building persistent AI visibility across a fragmented landscape.
Study Highlights
The Citation Persistence Problem
The average Citation Retention Rate across all five AI platforms over 28 days was 33%. That means two-thirds of the URLs cited in any given wave were replaced by entirely different sources within four weeks. Gemini and AI Overviews were 100% volatile. Not a single query held its sources steady over 28 days on either platform. Only 119 of the 1,127 unique URLs we catalogued appeared in all three collection waves, a survival rate of 10.6%.
AI visibility is not a position you earn once. It is a position you hold, lose, and re-earn on a cycle measured in weeks.
60% of AI Citations Are Invisible to Traditional SEO Tools
We compared every domain cited by an AI platform against the top 20 organic results on Google and Bing for that same query. Across all three waves, 57% to 62% of AI-cited domains did not appear anywhere in those organic results. We call this “dark matter” because it represents a citation layer that traditional rank trackers, SERP monitors, and SEO dashboards cannot see. The rate held steady across all three waves, which means this is not a transient anomaly. It is structural to how AI engines source their answers.
If your visibility strategy is built entirely on organic SERP performance, you are blind to the majority of the AI citation landscape.
Every AI Platform Is a Different Game
The domain overlap between any two AI platforms is extremely low. The highest Jaccard similarity we measured was 17% (Perplexity and AI Overviews). Most platform pairs shared fewer than 10% of cited domains. Between 78% and 85% of all cited domains were unique to a single platform.
ChatGPT cited sources only 26% of the time, relying on training data and brand entity recognition for the other 74% of responses. Perplexity averaged 5.9 citations per query with only 1 zero-citation response out of 90 data points, making it the most transparent engine in our study. AI Overviews drew heavily from YouTube, with 47% to 63% of responses citing YouTube content across our three waves. Copilot pulled from deep in Bing’s index, citing marketplace listings and niche domains that no other platform referenced. Gemini had the lowest citation retention of any platform at 11%, and our team documented it tightening its safety disclaimers in real time between Wave 1 and Wave 3.
Optimizing for one AI engine does almost nothing for the others. A platform-specific strategy is not optional.
Your Own Website Gets Cited 11% of the Time
Across recommendation and comparison queries where brands were actively mentioned, 89% of supporting citations pointed to third-party editorial sources. Only 11% pointed to the brand’s own domain. This held true even when the AI engine recommended the brand by name. The implication is clear. Getting featured in authoritative third-party roundups and editorial coverage may deliver more AI visibility than anything you do on your own website.
AI Engines Go Quiet on Sensitive Topics
Queries classified as Your Money or Your Life (YMYL), covering health, legal, and financial topics, showed systematically lower citation rates across every platform. ChatGPT returned zero citations on 97% of YMYL queries. Gemini returned zero citations on 73%. AI engines answered these questions confidently and often mentioned brands by name, but pointed to almost no verifiable sources. Legal services was structurally the most challenging vertical in our study, with a 24% average Citation Retention Rate and 94% of query-platform pairs classified as volatile.
Query Type Determines Brand Visibility
The type of query a person asks determines whether AI mentions a brand at all. “Best” and recommendation queries produced brand mentions 89% of the time. Comparison queries dropped to 40%. Conversational queries, the ones phrased the way people actually talk to AI chatbots, dropped further to 36%. Definitional queries (“What is...”) produced brand mentions only 3% of the time. If your AI visibility strategy is built around definitional content, this data shows that investment has a near-zero return for brand recognition.
Peek Inside the Guide
Table of contents
Understand the citation persistence problem.
See the full CRR data across all five platforms and five intent types.
Learn why 87% of query-platform pairs fall below the stability threshold and what complete citation turnover looks like in practice.
Explore the dark matter phenomenon: 60% of AI citations from domains invisible to traditional SEO tools.
See how Google’s #1 organic result performs across AI platforms (ChatGPT cites it only 9% of the time).
Examine the gap between SERP rank and AI citation frequency.
Review the full platform fragmentation analysis.
See ChatGPT’s memory-based recommendation model, AI Overviews’ YouTube dependency, Copilot’s unique sourcing behavior, Perplexity’s transparency advantage, and Gemini’s mid-study behavior changes.
Map brand visibility rates by query type.
Identify which intent categories produce citations and which produce source-free answers.
See why YMYL queries face systematic citation suppression and which verticals are structurally the most difficult.
Access the AI Visibility Stack: a six-layer framework for building persistent visibility across platforms.
Get vertical-specific platform priority recommendations, query-type action plans, a 30-60-90 day roadmap, and a self-assessment diagnostic to identify your blind spots.
Similar white papers
Parents don’t search for ABA therapy randomly. They search in patterns tied to diagnosis timelines, school calendars, insurance enrollment windows, and awareness cycles. A parent...
Google classifies every piece of legal content as “Your Money or Your Life.” That means your practice area pages, your attorney bios, and your blog...
Law firms cited inside Google AI Overviews earn 35% more organic clicks than firms at the same ranking position without a citation. AI Overviews now...
Download the Full AI Visibility Gap Study
Free download. No sales call required.