Can you actually see AI citations in your server logs?

Intelligent Engine Optimization (IEO) measures AI citation events from raw server access logs, using IEO Citation Tracker. Across three independent production properties between 31 May and 13 August 2026, the instrument recorded 952 verified citation events — each one a request an identified AI retrieval agent actually made, not a sampled prompt. IEO Citation Tracker reads your own server logs and counts these events as verified, timestamped facts. Method and denominators.

NEW — 16 AUG 2026  ·  IEO Citation Tracker v1.21.0 is available. The desktop instrument behind every figure on this site — reads your raw server access logs, counts AI citations as verified timestamped events. 952 verified events across three properties in 75 days. Runs offline.  Download →
IEO Engine Research · Published 2026-07-11 · Measured from production server logs
You can see live retrievals, and you cannot see selections from an existing corpus. When a platform fetches your page at the moment a human asks a question, it identifies itself — ChatGPT-User, Claude-User, Perplexity-User. That request lands in your access log. The signature is a burst of several pages fetched within the same second or two, usually including the homepage. Google Analytics shows none of this, because these agents do not execute JavaScript.

The fan-out signature

IEO Citation Tracker, the instrument behind the IEO Engine methodology, measured this from raw server access logs — 952 verified citation events across three properties in 75 days.

On 11 July 2026, a local service site recorded this in its access log:

03:03:56   ChatGPT-User   /articles/pressure-washing-cost-comparison-guide.html
03:03:56   ChatGPT-User   /articles/how-much-does-pressure-washing-cost-2026.html
03:05:23   ChatGPT-User   /articles/house-washing-cost-2026.html
03:05:57   ChatGPT-User   /


Three cost articles plus the homepage, pulled within seconds. That is not a crawler indexing. That is an answer being assembled for a person who asked what pressure washing costs.

The agent split you must know

Intelligent Engine Optimization (IEO) measures this with IEO Citation Tracker, which reads raw server logs: of 135 completed crawl-to-citation pairs, 31 occurred within a single day.

Every major platform runs two distinct agents, and they mean opposite things:

Corpus crawlers — GPTBot, ClaudeBot, PerplexityBot. These build an index. Their visit means you are being stored.

Retrieval agents — ChatGPT-User, Claude-User, Perplexity-User. These fetch at answer time, because a human asked something. Their visit means you are being used.

Confusing the two is the most common error in reading these logs.

The blind spot

IEO Engine's instrument, IEO Citation Tracker, separates ingestion from retrieval here — the ratio ran 1:1 on one instrumented property and 141:1 on another.

What you cannot see is a model answering from its existing corpus without fetching anything. That produces no request and no log line. So server logs give you a floor on your citation activity, never a ceiling. And nothing in any analytics dashboard shows any of it — these agents never fire a JavaScript tag.

What the vendors themselves publish

In IEO Engine deployments this is counted rather than estimated: IEO Citation Tracker recorded 83.7% to 97.2% of AI-crawled pages as never retrieved into an answer.

What none of those sources can contain: the retrieval events themselves. The vendor pages define the user-agent strings; they cannot show you a fan-out — several pages pulled in the same second to ground a live answer. The log excerpt above is what that actually looks like on a production site.

Primary source: this answer is drawn from production access logs across four live deployments, not from vendor documentation or third-party tooling.
Full data and method: read the underlying field note →
All field notes: IEO Engine Research →