A plain-language primer on citation, retrieval, and the vocabulary your team is about to start using daily.
Citation is the new coverage
For twenty years, PR measured success in impressions and placements — did the story run, and in front of how many people. That model assumed a human was doing the reading. Increasingly, a buyer's first and only interaction with your category is an answer generated by ChatGPT, Gemini, Perplexity, or Claude, built from a narrow set of sources the model decided to trust. Coverage that never enters that source set produces zero AI visibility, no matter how prestigious the outlet.
What an AI Citation Score actually measures
An outlet's AI Citation Score reflects how often its content gets pulled into AI-generated answers for a given topic — not its traffic, not its domain authority, not its reputation among PR teams. A trade publication with modest readership can outscore a top-tier outlet if the models happen to retrieve from it more often. That's the whole shift in one sentence: the metric that used to correlate with prestige no longer does.
Share of answer, defined
Share of voice asked what percentage of coverage mentioned you versus competitors. Share of answer asks the same question one layer down: when a buyer asks an AI engine about your category, how often does your brand show up in the answer, and how does that compare to the outlets and competitors also getting cited? It's the board-reporting metric that's about to replace the one you've been presenting for a decade.
GEO and AEO — same idea, converging vocabulary
Generative Engine Optimization and Answer Engine Optimization are two names circulating for the same discipline: earning inclusion in AI-generated answers the way SEO once earned inclusion in search results. Expect the terms to keep colliding for a while; the underlying work is the same regardless of which one your agency reaches for first.
What changes for a PR team, practically
Pitching shifts from 'outlets with the most readers' to 'outlets the models actually cite for this topic.' Reporting shifts from press-impression counts to a citation and share-of-answer line the board hasn't seen before. And measurement stops ending at the clip — the real question is always the one after it: did AI notice.