Findings · measured, dated, recomputable

52,413 verified ingestion fetches across 52,333 distinct URLs produced 5 verified live-answer fetches.

When an AI answer engine crawls a site thousands of times, how often does it actually cite that site to a human?

Cite as
An answer engine ingested 52,413 pages across 52,333 distinct URLs over 20 days and performed 5 verified live-answer retrievals — a ratio of 10483:1 (Still OS Digital Holdings LLC, answer-engine citation ledger).

The numbers

Verified ingestion fetches52,413
Distinct URLs ingested52,333
Days observed20
Verified live-answer retrievals5
Ingest : live-answer10483 : 1
Search-bot identity claims that were forged3 of 7 (42.9%)

Why it matters

Ingestion volume is not citation. These observations indicate that a site can be read exhaustively and continuously by an answer engine while almost never being retrieved to answer a person — and that crawler identity asserted in a user-agent string is sometimes asserted by something else entirely.

Method

Every request is matched to the crawler operator's OWN PUBLISHED IP RANGES (openai.com/gptbot.json, chatgpt-user.json, searchbot.json). A user-agent claim from an out-of-range address is counted as forged, never as a visit. Ingestion crawlers and live-answer retrievals are counted separately because only the second indicates a person asked a question.

Producer

The files these figures were derived from, on the machine that published this page

/home/marcus/state/agent-visitor-log.jsonl
/home/marcus/state/aeo/openai-gptbot.json

Next

How we measure it

Built 2026-10-30T22:17:03.185Z · JSON