Google Search Console Can Now Measure AI Search Visibility. Here's What Businesses Should Actually Track

Google finally gave website owners something the SEO industry has spent the last two years pretending it already had: a first-party view into whether a site is actually appearing inside Google's generative AI search features.

In June 2026, Google announced a dedicated Generative AI performance report in Search Console. By August 31, Google said the reporting had rolled out worldwide.

That matters.

It also does not mean Search Console suddenly tells you every prompt somebody typed, every answer your company appeared in, or exactly why an AI system chose your page.

The useful part is narrower and, in my opinion, better: we finally have a first-party signal that can separate some of the conversation about AI visibility from the mountain of third-party guesses.

For businesses trying to understand SEO, AEO and GEO in 2026, that is a meaningful upgrade.

What Google Search Console actually measures now

Google's new Generative AI performance report shows visibility from supported generative AI features in Google Search, including AI Overviews and AI Mode.

The report currently gives website owners views such as:

• total generative AI impressions over time
• which pages received those impressions
• the countries those impressions came from
• device breakdowns
• date-based trends

Google defines an impression here as a time a link from your site was shown to a user in a supported generative AI feature.

That is important wording.

An impression does not automatically mean a user clicked you. It does not automatically mean your brand was the centerpiece of the answer. And it does not tell you the exact prompt that caused the appearance.

It tells you something much more defensible:

Google surfaced one of your URLs inside a supported generative AI search experience.

That is the kind of data businesses should be building decisions around.

Google's official announcement: Search Generative AI performance reports in Search Console

What the report does not tell you

This is where I expect a lot of bad dashboards to appear.

The new report is useful, but it is not a magical "AI ranking" score.

At the moment, it does not give you a clean list of every generative search prompt that exposed your website. It does not tell you that you are "the number three result in AI Mode." It does not reveal Google's internal ranking logic. It does not prove that a specific piece of schema markup caused an AI citation.

And it definitely does not justify inventing a proprietary GEO score and pretending it came from Google.

Google itself warns site owners to be careful with third-party tools that claim to have access to internal Google ranking or AI metrics.

That does not make outside visibility tools useless. We use external observation, testing and entity research because first-party reporting never tells the whole story.

But there is a difference between measuring observable signals and inventing certainty.

That distinction matters more as AI search becomes a real acquisition channel.

The five things I would actually track

If I were building an AI visibility dashboard for a business today, I would not start with one giant GEO score.

I would start with five buckets.

1. Generative AI impressions

This is the new first-party baseline.

Is Google showing your pages inside AI Overviews and AI Mode at all?

Then:

• is that visibility growing?
• which pages are earning it?
• does it concentrate around one topic?
• does it expand as the site develops more authority?

You cannot improve what you never establish a baseline for.

2. The pages earning AI visibility

This may be more useful than the headline impression number.

If the same three pages repeatedly earn visibility, that tells you something about what Google understands your business to have authority on.

That can guide:

• content expansion
• internal linking
• service-page improvements
• expert commentary
• original research
• supporting FAQs and definitions
• proof and corroboration

A business should be asking: what do our visible pages have in common?

Then build outward from that.

3. Traditional organic performance

AI visibility does not replace SEO.

Google's own guidance on generative AI search is clear that the fundamentals of being discoverable, crawlable and useful still matter.

So I still want:

• non-branded search impressions
• clicks
• click-through rate
• ranking movement
• indexed pages
• crawl health
• conversion paths

If a company is invisible in ordinary organic search, has thin service pages, weak entity clarity and no useful content, I would not expect a clever "GEO optimization" trick to rescue it.

That is why we treat SEO, AEO and GEO as one visibility system, not three unrelated services.

4. Branded and entity-level corroboration

AI systems do not learn what a company is only from the company's own website.

They can encounter the business across:

• Google Business Profile
• industry directories
• news coverage
• social profiles
• review platforms
• professional associations
• other websites
• structured business data
• public references to people, products and services

This is the part of GEO conversations that often gets reduced to "get mentioned on Reddit" or "add schema."

The real issue is broader: does the web tell a consistent story about who you are, what you do, where you operate and why you should be trusted?

That is entity clarity.

For local and real estate businesses, it is especially important because location, expertise, service area and reputation are part of the question being answered.

Our deeper breakdown for agents is here: Generative Engine Optimization for Real Estate Agents.

5. Business outcomes

This is the one people will skip because impressions are easier to screenshot.

AI visibility only matters if it eventually contributes to something useful.

Depending on the business, that might be:

• qualified leads
• direct brand searches
• contact-form submissions
• calls
• demo requests
• booked appointments
• assisted conversions
• sales conversations

Attribution will remain imperfect.

Someone can encounter your business in an AI answer, remember the name, search you later and convert through a branded result.

The answer is not to pretend attribution is perfect. The answer is to combine first-party visibility, search behavior and business outcomes and look for movement over time.

This is why "GEO rankings" have always bothered me

Traditional search results at least gave the industry a familiar concept: position.

Even that metric can be misleading because rankings vary by location, device, intent and result type.

Generative search is even less compatible with a simple leaderboard.

An AI answer can synthesize several sources. It can reformulate the user's question. It can expose different links or supporting material depending on how the conversation develops.

Trying to compress all of that into "you rank #4 in ChatGPT" is usually more confidence than the data deserves.

I would rather tell a client:

Google is now showing these five pages in generative search. Visibility increased. These two topic clusters are doing most of the work. Branded search is also increasing. Here is what we are building next.

That is actionable.

And now at least part of it can come directly from Google.

The newest signal is multimodal search

There is another reason businesses should stop thinking about SEO as ten blue links.

On September 24, 2026, Google added reporting for web multimodal search to both traditional Search performance and the Generative AI performance report.

That means site owners can begin seeing when their web content surfaces from searches initiated with images, including experiences such as Google Lens, Circle to Search, uploaded images and Chrome image search.

For businesses with visually important inventory, places, products, properties or projects, that deserves attention.

A real estate company, for example, should not think of property photography as separate from search visibility anymore.

A contractor should not assume project imagery is only portfolio decoration.

A product company should not treat image optimization as a 2014-era alt-text checklist.

The search interface itself is becoming multimodal.

Google's announcement is here: web multimodal Search performance reporting.

What I would do inside Search Console this week

If you own or manage a website, I would do this before buying another GEO tool.

Open Search Console and look for the Generative AI performance report.

If your property has enough qualifying visibility to populate the report, capture a baseline:

1. total generative AI impressions
2. top pages by generative AI impressions
3. country distribution
4. device distribution
5. trend by date

Then compare the visible pages against your normal Search performance.

Ask:

• Are our AI-visible pages also strong traditional organic pages?
• Are informational pages doing better than commercial pages?
• Which topics does Google appear willing to surface us for?
• Do those topics match what we actually want to be known for?
• Are there service pages that should be supported by stronger explanatory content?
• Are the pages internally connected to the next useful action?

If the generative report is empty, that is also information.

It does not mean the company is doomed. It means you do not yet have enough supported generative AI impressions for Google to populate the report.

That is a much better starting point than pretending otherwise.

What MetaKona is tracking

We are treating this the same way we treat the rest of search: as an evidence problem.

For MetaKona's own site, Search Console is already testing pages across topics including MCP for small business, real-estate GEO, IDX and AI/search strategy.

That does not mean we "won GEO."

It means Google has started associating the site with those topics and is testing whether our pages deserve more visibility.

The job now is to strengthen the clusters that are showing signs of life, improve the pages that are sitting within striking distance, create better internal connections between them and keep publishing information based on work we are actually doing.

That last part matters.

The easiest way to create generic AI-search content is to ask AI to write about AI search.

There is already enough of that.

What businesses need is evidence, point of view, first-party experience and pages that answer real questions better than the pile of interchangeable summaries already online.

That is what SEO was supposed to be anyway.

SEO, AEO and GEO are converging

The more search changes, the less useful it becomes to pretend SEO, AEO and GEO live in separate boxes.

Good technical SEO helps search systems discover and understand your site.

Good content gives those systems something worth retrieving.

Clear entities and corroboration help them understand who the information belongs to and whether it can be trusted.

Strong answer structure makes important information easier to extract.

Original evidence gives a page something that cannot be recreated by scraping ten other articles.

Internal links establish relationships between topics.

Conversion design turns visibility into business.

That is not three strategies.

It is one modern visibility system.

Google finally giving site owners a first-party window into generative AI impressions makes that system easier to measure.

It does not make it simpler to fake.

And that is probably a good thing.