Questions
Frequently asked questions
What AgentReady measures, what it refuses to measure, and where each number stops being reliable.
What is AI readiness?
- Whether an AI crawler can fetch your pages and parse what is on them. It is measured deterministically across seven categories and scored out of 100. It is a separate question from whether engines cite you, and the two are never combined into one number.
Why can a page rank on Google and still be invisible to ChatGPT?
- Retrieval crawlers do not run JavaScript. Googlebot renders pages; OAI-SearchBot and PerplexityBot read the HTML they are served. A site whose content only exists after hydration is blank to them, which is why rendering carries 30 of the 100 readiness points.
Does blocking GPTBot stop ChatGPT from citing me?
- No. GPTBot is a training crawler; OAI-SearchBot builds the index that ChatGPT Search draws answers from. They are different user agents and different decisions. Blocking the training tier is also not retroactive — it cannot remove you from a model already trained.
What is llms.txt and do I need one?
- A plain-text index at /llms.txt that tells a model what a site contains and where. It is worth 5 of 100 readiness points here — useful, not decisive. A stale one is worse than none, because it asserts a structure the site no longer has.
Why is there no single AI visibility score?
- Because averaging across engines discards the signal. Only about 11% of domains are cited by both ChatGPT and Perplexity, so a blended number hides the very difference you need to act on. Rates are reported per engine, each with its sample size and interval.
How is citation rate measured?
- By sampling: the same prompts are run repeatedly against each engine and the results are counted. Identical prompts vary 10–34% between runs, so every rate ships with a 95% Wilson interval and its sample size. Below 100 runs the run count is shown instead of a percentage.
Is readiness the same as being cited by AI?
- No, and they are never combined here. Readiness is deterministic — can a crawler fetch and parse the page. Citation is sampled and probabilistic. A perfectly readable site with nothing worth citing scores high on one and zero on the other.
How long before a fix shows up in the numbers?
- A fix is judged after a 21-day window, not on deploy day. Crawlers revisit on their own schedule and citation rates are noisy between runs, so a same-day reading measures variance rather than the change.
Why are AI referral counts called a lower bound?
- Because a large share of AI traffic arrives with no referrer and lands in Direct. Analytics can only count what identifies itself, so the real figure is higher by an amount that cannot be determined for any specific site.
What does a missing day on a rank chart mean?
- That no impressions were reported, not that rank fell to zero. Search Console and Bing Webmaster only report queries a site already appears for, so an absent day is drawn as a gap and never as a drop.