September 16, 2026 · 6 min read

What is the Search Console AI performance report actually telling us?

In The Report:
NO CLICK DATA

.smk-article { –ink: #FAF8F4; –ink2: #FFFFFF; –paper: #FFFFFF; –teal: #1C8A7D; –blue: #2F5FD6; –gold: #A9701F; –line: rgba(15,23,32,0.10); –muted: #6B7785; –muted2: #6B7785; }
.smk-article * { box-sizing: border-box; }
.smk-article { scroll-behavior: smooth; }
.smk-article { margin: 0; background: var(–ink); color: #0F1720; font-family: Inter, sans-serif; line-height: 1.6; -webkit-font-smoothing: antialiased; }
.smk-article main { max-width: 760px; margin: 0 auto; padding: 8px 24px 40px; }
.smk-article .seclabel { font-family: “JetBrains Mono”, monospace; color: var(–teal); font-size: 0.85rem; letter-spacing: 0.08em; margin: 56px 0 10px; }
.smk-article h2 { font-family: Fraunces, serif; font-weight: 500; font-size: clamp(1.5rem, 3vw, 2.1rem); line-height: 1.2; margin: 0 0 20px; }
.smk-article p { color: var(–muted); margin: 0 0 18px; }
.smk-article .stats { display: grid; grid-template-columns: repeat(auto-fit, minmax(150px, 1fr)); gap: 1px; background: var(–line); border: 1px solid var(–line); border-radius: 14px; overflow: hidden; margin: 28px 0 40px; }
.smk-article .stat { background: var(–ink2); padding: 22px 18px; }
.smk-article .stat .num { font-family: “JetBrains Mono”, monospace; font-size: 1.9rem; color: #0F1720; border-top: 2px solid var(–teal); padding-top: 12px; display: inline-block; }
.smk-article .stat.b .num { border-color: var(–blue); }
.smk-article .stat.g .num { border-color: var(–gold); }
.smk-article .stat .lbl { font-size: 0.82rem; color: var(–muted); margin-top: 8px; line-height: 1.4; }
.smk-article .stat .src { font-family: “JetBrains Mono”, monospace; font-size: 0.66rem; letter-spacing: 0.06em; text-transform: uppercase; color: var(–muted2); margin-top: 10px; }
.smk-article figure { margin: 44px 0; padding: 28px 24px 20px; border: 1px solid var(–line); border-radius: 16px; background: #FFFFFF; }
.smk-article figure svg { width: 100%; height: auto; display: block; }
.smk-article figcaption { font-family: “JetBrains Mono”, monospace; font-size: 0.68rem; letter-spacing: 0.07em; text-transform: uppercase; color: var(–muted2); text-align: center; margin-top: 16px; }
.smk-article blockquote.pull { border-left: 3px solid var(–teal); padding: 4px 0 4px 26px; margin: 48px 0; font-family: Fraunces, serif; font-style: italic; font-size: 1.35rem; line-height: 1.5; color: #0F1720; }
.smk-article .cards { display: grid; grid-template-columns: repeat(auto-fit, minmax(220px, 1fr)); gap: 18px; margin: 28px 0 40px; }
.smk-article .card { background: #FFFFFF; border: 1px solid var(–line); border-radius: 14px; padding: 24px; }
.smk-article .card .num { font-family: “JetBrains Mono”, monospace; color: var(–teal); font-size: 0.78rem; letter-spacing: 0.08em; margin-bottom: 10px; }
.smk-article .card h3 { font-family: Fraunces, serif; font-weight: 500; font-size: 1.1rem; margin: 0 0 10px; }
.smk-article .card p { font-size: 0.92rem; margin: 0; color: var(–muted); }

{
“@context”: “https://schema.org”,
“@type”: “BlogPosting”,
“@id”: “https://sentientmarketer.com/signal/search-console-ai-performance-report/”,
“headline”: “What is the Search Console AI performance report actually telling us?”,
“description”: “Rolled out to every site on 31 August 2026, the report gives impressions and no clicks. That changes what a CEO can safely conclude from it.”,
“author”: {
“@type”: “Person”,
“name”: “David Fox Jensen”,
“jobTitle”: “Founder & Principal”,
“url”: “https://sentientmarketer.com/”
},
“publisher”: {
“@type”: “Organization”,
“name”: “The Sentient Marketer”,
“url”: “https://sentientmarketer.com/”
},
“datePublished”: “2026-09-16”,
“dateModified”: “2026-09-16”
}

It is telling you where you showed up, and almost nothing about what showing up was worth. Google finished rolling the generative AI performance report out to every site in Search Console on 31 August 2026, after a phased launch that started in June, and the report carries impressions only: no clicks, no click-through rate, no average position, and no view of the prompts that produced the appearance (Google Search Central Blog, June 2026; Search Engine Land, August 2026). So you now have a line that moves, attached to a surface where Pew Research found people click a traditional result in 8% of searches that carry an AI summary against 15% of searches that do not.

That gap is the whole story of the report. A rising impression count on a surface with a falling click rate can mean you are winning, or it can mean the summary is answering the question so completely that nobody needs you. The report cannot tell the difference, and neither can anyone who reads it to you without saying so.

/ 01

What is actually inside the report

The generative AI performance report separates out your impressions in Google’s generative surfaces, which means AI Overviews and AI Mode in Search, plus generative features in Discover. You can break those impressions down by page, country, device and date, and you can watch the trend over time. That is the full menu.

What is missing matters more than what is present. There are no clicks, so there is no click-through rate and no conversion path. There is no query or prompt data, so you cannot see what someone asked to make you appear. There is no position, so you cannot tell whether you were the first source named or the ninth. Google said at launch that it would add metrics over time, and in the months since it has publicly acknowledged that the current reporting does not give site owners what they need (Search Engine Journal, 2026). Treat that acknowledgement as a dating stamp on everything the report currently says.

There is also a definitional wrinkle worth knowing before you set a target on this number. An impression records that your link was present in the generated answer. It does not record that the reader saw it. Links tucked behind a “Show more” control are counted, and a citation a reader never expanded is worth a great deal less than one sitting in the opening sentence. The report flattens both into the same unit.

31 AUG
Date the AI performance report reached every site worldwide, after a June phased launch
Search Engine Land, August 2026
0
Click, click-through rate and position metrics in the report at launch
Google Search Central Blog, June 2026
8%
Of searches with an AI summary produced a click to a traditional result, against 15% without one
Pew Research Center, July 2025
1%
Of visits involved clicking a link inside the AI summary itself
Pew Research Center, 68,879 searches

WHAT A CLICK IS WORTH ON EACH SURFACE

NO AI SUMMARY
AI SUMMARY SHOWN
LINK IN SUMMARY

15%
8%
1%

SHARE OF SEARCHES PRODUCING A CLICK

Source: Pew Research Center analysis of 68,879 Google searches from 900 US adults, July 2025
/ 02

Why an impression count with no click count gets misread

Scale is about to make this number look impressive. AI Overviews now reach more than 2.5 billion monthly users and AI Mode has passed a billion, so almost any site with reasonable topical coverage will watch its generative impressions climb through the rest of the year. The climb will be real. The conclusion most boardrooms will draw from it will not be.

The misreading works like this. A marketing team opens the report, sees impressions up 40% quarter on quarter, and presents it as evidence that the AI visibility work is paying off. Nobody in the room asks what the same chart would look like if the company had done nothing at all, because the report offers no benchmark, no competitor set and no share figure. The number goes up for everyone as the surface grows. Movement is not proof of skill when the tide is coming in.

The opposite misreading is just as common and more expensive. A CEO sees flat traffic alongside rising AI impressions and concludes the whole AI visibility question is noise. That ignores what the citation is doing while the click is not happening. Being named in the answer shapes the shortlist a buyer arrives with, which is the part of the funnel your analytics has never been able to see and never will.

An impression is a presence metric. Treating it as a demand metric is how a company spends two years getting better at being mentioned and worse at being chosen.

Pew’s underlying behavioural data is the clearest available guide to what to expect. Readers clicked a link inside the summary itself in 1% of visits. They were more likely to abandon browsing entirely after a page with a summary, 26% of the time against 16% without. And longer, more conversational queries were far likelier to trigger a summary at all, with 8% of one and two word searches producing one against 53% of searches of ten words or more. The queries where a mid-market buyer describes their actual problem are exactly the queries where the summary appears and the click does not.

None of this makes the report worthless. It makes it a share-of-voice instrument rather than a traffic instrument, and share of voice has always been read against competitors and over long windows, never as a monthly movement in isolation.

WHAT THE REPORT SEES AND WHAT IT DOES NOT

REPORTED
ABSENT

IMPRESSIONS
DATES
DEVICE
COUNTRY
CLICKS
POSITION
PROMPTS
CTR

Sources: Google Search Central Blog, June 2026; Search Console Help, generative AI performance report documentation
/ 03

What you can actually do with it this quarter

01

Set the baseline before you set a target

Export the last available 90 days by page and country and freeze it. Every future claim about AI visibility at your company should be measured against a number that existed before anyone had an incentive to move it.

02

Pair it with the two numbers it cannot give you

Branded search volume and a self-reported “how did you hear about us” field on your forms. Impressions tell you the machine noticed. Those two tell you whether a human did.

03

Sort by page and look for the unmonetisable winners

Pages that earn heavy generative impressions and nothing else are answering a question you do not get paid for. That is useful information about what to write next, and about what to stop writing.

04

Ban the standalone slide

The number never appears in a board pack on its own. It appears next to pipeline, branded search and win rate, or it does not appear. A metric with no click attached cannot carry a page by itself.

05

Re-read it in six months

Google has committed to adding metrics. When clicks arrive, every conclusion drawn from the impressions-only era becomes testable, so keep a written record of what you concluded and why.

/ 04

The part nobody wants to hear

Most companies will add this report to the monthly deck and change nothing. The report will become a slide, the slide will become a habit, and the habit will feel like progress for about three quarters. This is the same thing that happened to bounce rate, to impressions share, to the social follower count, and to every other number that was easy to obtain and hard to act on.

The reason is structural rather than intellectual. Nobody in a mid-market company owns the question the report raises, which is whether being cited by a machine is changing who ends up on the shortlist. Answering it requires connecting a Search Console export to a CRM field to a set of sales conversations, and that work belongs to whoever has ten uninterrupted hours to do it. In most companies that person does not exist, so the work does not happen, and a new dashboard arrives instead. Capability purchased first, question defined later, if at all.

The companies that will get something out of this report are the ones that decided, before the report existed, what they wanted to be the answer to. For them the impression count is confirmation or correction. For everyone else it is a number with nowhere to go.

Read the AI performance report as share of voice on a surface you do not control, and never as traffic you earned.

SOURCES: GOOGLE SEARCH CENTRAL BLOG, GENERATIVE AI PERFORMANCE REPORTS, JUNE 2026 · SEARCH ENGINE LAND, GLOBAL ROLLOUT, AUGUST 2026 · PEW RESEARCH CENTER, 68,879 GOOGLE SEARCHES, JULY 2025 · SEARCH ENGINE JOURNAL, GOOGLE ON REPORTING GAPS, 2026 · GOOGLE SEARCH CONSOLE HELP, GENERATIVE AI PERFORMANCE REPORT
DFJ
David Fox Jensen
Founder & Principal · The Sentient Marketer

20+ years leading marketing for 150+ brands. Now runs a principal-led fractional CMO and AI consulting practice for mid-market companies. Writes The Signal weekly on what the machines are learning about your business. Machines for scale, never for soul.

Subscribe

Wondering what the machines say about you?