Your buyer's agent just tried to read your pricing page

Your next buyer might never load your site. Their agent will. It fetches the pricing page, reads the docs, and reports back. Whether it can read anything on your domain is now a pipeline question.

Vercel and the ora.ai research lab just published the data: 1,033 agent runs across 25 sites, plus 190 controlled fetch probes. Three findings worth your attention.

One. Agents read the same pages people do. 82.5% of answers traced back to a page the agent actually fetched. Of those, 47% came from a docs page. Not llms.txt. Not structured data. Plain documentation.

Two. Only two things broke retrieval in the controlled probes. Answers hidden behind JavaScript: the page ships an empty shell and fills in content after load, so a fetch-only agent reads nothing. And bot blockers returning 403 to agent traffic. Every one of the other 16 configurations passed.

Three. The fancy stuff didn't move the needle. Adding llms.txt, JSON-LD, sitemaps, and Markdown mirrors changed nothing when the answer was already in the server-rendered HTML. llms.txt steered agents to a listed page in 36% of runs that reached it. Useful. Not a substitute for readable pages.

The business translation: if your pricing sits behind an app shell, or your bot blocker treats every agent as a scraper, the agent answers your buyer's question from someone else's page. Usually a competitor's comparison post. You lose the deal without ever seeing the traffic.

One honest caveat before you quote these numbers in a Slack thread: six sites account for 67% of the runs. Treat the percentages as this sample, not the web.

The check costs nothing. Paste your domain into is-agentic.com and you get a scored report. Full guide:

https://vercel.com/kb/guide/make-your-site-readable-by-ai-agents

Post your score below. What blocked you? My money's on the bot blocker.