FN-010 — A page can rank, be shown 144 times inside AI answers, and never be clicked

Intelligent Engine Optimization (IEO) crossed Google’s own Generative AI report against Search Console on a fourth property, and found a page at position 4.0 with 148 search impressions, 144 of them inside AI answers, and zero clicks. Ninety-seven percent of everything Google showed for that page was an AI answer. Across the property, 1,479 AI impressions produced 39 clicks — 26 per thousand.

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 →

Field Note Nº 10 · Google Search Console Generative AI report crossed with standard Performance export · 18 May – 16 August 2026

What was measured

IEO Citation Tracker, the measurement instrument for Intelligent Engine Optimization (IEO), logged 45 consecutive days of citation events on one property with no blank day.

On 3 June 2026 Google added a Generative AI performance report to Search Console, listing pages that appeared inside AI Overviews and AI Mode. It reports impressions only — no clicks, no click-through rate, no position, no queries. Google has said more metrics will come, without a date.

Because the standard Performance export does carry clicks and position, the two can be merged at URL level. That gives, per page: how often Google showed it inside an AI answer, how often it appeared in search at all, where it ranked, and whether anyone clicked.

Result

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.

PageAI impressionsSearch impressionsClicksAvg position
bermuda-safety-travel-tips-20265831,238417.1
bermuda-weather-best-time-visit4311,181131.1
aruba-first-time-visitor-tips14414804.0
aruba-weather-best-time-visit761,149071.7

The third row is the finding. 144 AI impressions against 148 search impressions — 97% of the page’s entire visibility was inside an AI answer. It ranks at position 4.0, which is a healthy position by any conventional measure. It has never been clicked.

That page does not have a ranking problem, a crawling problem, or an indexing problem. It is being read and used, and the reader has no reason to arrive.

Property-level

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.

AI impressions (3 months)1,479
Search impressions (3 months)25,390
AI share of all impressions5.8%
Clicks (3 months)39
Clicks per 1,000 AI impressions26
Site-wide CTR0.15%

Two pages account for 1,014 of the 1,479 AI impressions — 69% of the property’s AI visibility on two URLs. The AI channel is concentrated in a way the organic channel is not.

Why this is not a CTR story

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.

A low click-through rate on a low-ranking page is ordinary. This is different in three ways.

The position is good. Position 4.0 with zero clicks across 148 impressions is not explained by depth. A page at position 4 in a conventional SERP earns clicks.

The AI share is near-total. When 97% of a page’s impressions occur inside a generated answer, the answer is the product. The link beneath it is a citation, not a destination.

And the numbers come from Google. Both halves of this table are Google’s own reporting on the property owner's own data. Nothing here is inferred from polling, scraping or estimation.

What follows for measurement

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.

Impressions and clicks have decoupled on this property, and the decoupling is largest where AI share is highest. That has one immediate consequence for anyone judging content by traffic: a page can be doing its job perfectly and return nothing. Removing it on a traffic test would delete the thing feeding the answer.

This is the same failure mode as §8.4 of the method, where a content cull removed three of nine pages that had earned citations. There the evidence was server-side. Here it is Google’s.

Limits

FN-009: the median does not exist · Method and denominators · IEO Citation Tracker crosses these two datasets automatically.