PR’s AI Visibility Numbers Are Fabricated Nonsense—And Everyone’s Quoting Them Anyway

Let’s kill this 84% myth before it infects another Monday-morning client meeting: No, your brand’s AI visibility is not ‘84% dependent on earned media.’ That number is pure fiction, cooked up by agencies who wouldn’t know a vector database from a press kit. Walk down Broadway at noon and you’ll hear some poor comms director parroting this stat as gospel, drowning in LinkedIn slides about ‘AI readiness’ and ‘media footprint.’

Here’s the ugly truth: AI engines, whether Gemini, GPT-5, or whatever OpenAI’s selling this week, do not build answers from press releases and PR hits alone. They pull from Wikipedia, Reddit, product docs, customer reviews, and a thousand other sources your PR dashboard can’t see. Counting ‘press coverage’ as the main driver of AI visibility is like measuring New York’s nightlife by the number of Times Square billboards—it’s a joke, and a lazy one at that.

David Clare from Fire on the Hill had the guts to call this out, and he’s dead right: if we expect brands to trust us on AI, we have to stop faking numbers. But the PR industry would rather churn out slide decks with fake stats than admit they’re guessing. It’s a cottage industry of self-congratulation, fueled by ‘AI visibility audits’ that are just glorified media clippings with an LLM sticker on top. This is peak nothingburger territory.

If you’re a CMO and your agency drops the 84% stat, ask them to prove it. Demand a real breakdown of your brand’s presence across sources that LLMs actually crawl—structured data, user forums, product documentation. If they can’t, cut them loose. AI visibility is earned, not spun. The PR industry needs to get out of its own echo chamber and start holding itself to the same standards it claims to set for everyone else.