A working list of the statistics we consider load-bearing for anyone thinking about AI visibility, each with its source. We update it as better data appears; last updated August 2026. Quote freely with attribution.

How big AI search already is

  • Around 800 million people use ChatGPT weekly as of 2026, with Google's AI Overviews reaching over 2 billion users monthly (company statements collected by Search Engine Land).
  • Gartner projects a 25 percent drop in traditional search volume by 2026 as conversational answers absorb queries (Gartner, 2024 projection).
  • Over 100 million people search with AI every day, per Profound's own marketing, which is notable as the number an enterprise vendor stakes its category on.

How few businesses actually show up

  • ChatGPT surfaces only about 1.2 percent of local businesses in local-recommendation answers, against roughly 36 percent surfaced by Google local results (analysis reported by Cognizo, 2026). The gap is the opportunity.
  • Only 6.8 percent of domains cited by ChatGPT also appear on three or more other AI platforms (cross-platform analysis via GetFancy, 2026). Visibility on one engine says little about the others, which is why single-engine tools mislead.
  • ChatGPT mentions brands about 3.2 times more often than it cites them (BrightEdge research). Being named and being used as a source are different achievements, and most businesses have neither.

What moves the answers

  • Brand mentions predict AI visibility at a correlation of 0.664, backlinks only 0.218, across a 75,000-brand study (Ahrefs, 2026). Being talked about beats being linked to, by a factor of three.
  • Adding statistics to a page improves its citation odds by roughly 41 percent; quotations and citations show similar lifts (Princeton, Georgia Tech and IIT Delhi GEO research, arXiv 2311.09735).
  • Comparison tables earn about 2.5 times more citations than average content, and brands with real Reddit or LinkedIn presence appear in answers up to 4 times more often (synthesis reported by GetFancy, 2026).

How unstable single measurements are

  • Two identical prompts return the same brand list less than 1 time in 100, and the same list in the same order about 1 time in 1,000, across a 2,961-run study (SparkToro, 2026).
  • 38 percent of identical prompts returned a meaningfully different brand list when asked three times (Vismore, 750-response audit, 2026).
  • The honest conclusion from both studies: AI visibility is a probability, not a rank. A brand appearing in 4 of 10 runs has a 40 percent mention rate; any tool reporting a single daily "position" is reporting noise. This is why Saymetry reports mention rates with confidence intervals and resamples a fifth of every scan to measure stability.

What it costs to know

  • Enterprise AI-visibility platforms run $89 to $499+ per month, with the best-known charging $99 monthly for a single engine and $399 for three (public pricing pages, August 2026).
  • A complete 8-engine Saymetry scan with a website audit and action plan is $99 once, which is the comparison we are happy to invite.