Co-citation gap analysis, the method Citation Labs described in Search Engine Land on June 22, reframes AEO work around a question most teams have not asked: when each person in a buyer committee queries an AI assistant about the same decision, which sources does the assistant cite for each role, and where does your content fail to appear?
The starting observation is a familiar one to anyone who has watched AI answers evolve. AI assistants do not return a single answer to a buying decision. They return role-sensitive answers. A CEO asking about a post-Series A rebrand gets different citations than in-house counsel asking about the same decision. The cited sources, the sub-queries the model generates, and the evidence the model treats as authoritative all shift by role.
The method works by holding the scenario constant and changing only who is asking. You write one first-person prompt per buyer role, run each against an AI tool that exposes its working, and capture three things: the sub-queries generated, the pages read, and the pages cited. The distinction between read and cited matters. Pages the model reads but does not cite represent a content problem; pages it never reads at all represent a discoverability problem. Both require different fixes.
From the raw data you build a citation matrix: each cited URL on its own row, each buyer role its own column. Sort by how many roles cite each source. The shape of the matrix tells you whether the committee’s evidence is convergent (a few shared sources) or disjointed (every role living in its own reference world). Most teams have been optimizing for the convergent core and ignoring the isolated seats.
The most valuable thing the matrix surfaces is what Citation Labs calls the veto-isolate seat: the role whose veto can kill the sale and whose citations share the least common ground with the rest of the committee. In the biotech logo example documented in the article, that role was in-house legal counsel, who returned 14 cited sources, not one of them overlapping another role, all drawn from trademark and regulatory domains. Lowest competition on the citation map. Highest leverage for a team willing to build there.
This framing matters for how AEO teams structure their planning cycles. Most content audits assess coverage by topic or keyword cluster. This method adds a second axis: buyer role. A piece of content can rank and appear in AI answers for one role while being entirely absent from another role’s answer on the same decision. The gap between those two states is not a ranking problem. It is a content-type problem, one that keyword analysis will not surface on its own.
The tactical output of the analysis is straightforward: the AI assistant’s own sub-queries tell you which domains it trusts per role. Those domains become your outreach list. The next content asset’s job is not to describe an option but to supply the evidence a specific role needs to reach a specific decision.
Any AEO team planning the next quarter’s content calendar should run this analysis on two or three buying decisions before briefing a single asset. The citation gaps it finds are more actionable than traffic projections.
Citation Labs authored this method and walkthrough, published by Search Engine Land on June 22, 2026.
