Search Console: measuring your visibility in AI responses

This content is part of our AI & Data series. At ELLEVATE, we help brands leverage AI and data without losing their vision or uniqueness. Here, you'll find concrete resources to intelligently integrate these levers into your strategy. See our data-driven supportOur marketing & management services

Google has (finally) announced the creation of a dedicated AI performance report in its Search Console. Note that this only covers appearances in its popular AI Overviews and AI Mode, the AI-powered response integrated into Google's search results page.

This is a first step towards monitoring AI performance! But of course, it's not all that simple.

The report primarily measures a site's exposure to Google's generative features. It does not provide a complete attribution between this exposure and leads or revenue.

In concrete terms, you still don't really know where the sale you just made came from and whether it was thanks to AI Overviews that it reached you.

What does the Generative AI Performance report measure?

The report allows us to observe the impressions associated with AI responses from Google search.

Google tells us that the data can be analyzed by page, country, device, and date. This includes AI Overviews and AI Mode, where these features are available for the site and market in question.

This means that you will not be able to get information about France if you search before the feature's deployment date in France.

This is a first step towards answering this question from Google's perspective: which pages of our site are actually integrated into Google's generative responses?

Why this data changes SEO reporting

Until now, much of the AI visibility had to be estimated from manual testing, third-party tools, or visits from referring domains associated with assistants.

The new Google report creates an intermediate layer between traditional SEO and conversion. It's now possible to track a web page as it appears on regular Google search results, and then in an AI chat.

To delve deeper into the AI visibility indicators ChatGPT, Claude, Perplexity and others, and the Share Of Model.

From there, it's a short step to analyze who generates the most attention towards your brand (brand searches), and the most direct purchases.

Let me give you an example:

October 6: One of your pages starts performing exceptionally well on Google's AI Overviews.
October 9th: You have a significant increase in your brand searches on Google
October 11: You receive a higher than usual number of incoming calls, emails, or discovery call bookings.

You can't actually link the high-performing AI page to your lead count. You don't have a data stream that clearly tells you that this page generated 70% more inbound leads.
However, by concording and comparing this information, you can hypothesize that page X played a significant role in the increase in your October revenue.

I've delved deeper into new ways to interpret your numbers and performance in this article about the new story your marketing KPIs are telling you today.

And the only way to confirm this is to optimize a page of the same type and analyze what happens.

What the report does not allow us to conclude

An AI printout does not prove that a user has viewed, remembered, or chosen your business.

The report alone does not allow us to know:

if the content was actually read
what wording of the answer was displayed
if your brand has been mentioned favorably
if the user continued their journey elsewhere and chose a competitor
if the visibility generated a business opportunity

It also doesn't cover all generative assistants and engines. ChatGPT, Claude, and Perplexity require different tracking methods. And given that each AI works differently, this is a crucial factor to consider.

However, it is one more indicator, and it multiplies the dotted lines along the path of your target customer to better understand where they go and how you are used by AI.

How to integrate AI visibility into the dashboard

I recommend 4 levels of reading here

1. The SEO Foundation

Track impressions, clicks, queries, and traditional landing pages. Because we're talking about Google, the generative visibility of AI Overviews remains linked to the quality of SEO, technical accessibility, and content.

What can be particularly interesting is to analyze which pages perform better in AI Overviews than in regular Google search and to identify areas of difference in optimization. Slightly boosting the optimization of a page that already performs well in AI could improve your Google ranking. Be careful not to over-optimize it, as this could cause you to lose AI impressions. Keep a close eye on this!

To track all of this in a specific and strategic way, we're implementing advanced GA4/GTM tracking systems for our clients. Since AI doesn't provide much information to begin with, we need to maximize what we can do with what we have!

2. Exposure in AI responses

Identify the pages exposed, their progression over time, and the markets concerned.

Analyze high-performing pages from two perspectives:

Pages with more commercial intent: those that sell you and recommend you as a trustworthy business
Informational pages: these are the pages used as a source for "how-to" type answers. These are the pages that build your authority, and they are essential for increasing your AI visibility.

If you want to build a GEO strategy that is not an outgrowth of your marketing or a complex system, but an element integrated into what you have already done, make an appointment for a discovery exchange.

3. Brand Demand

Observe branded queries, that is, those that specifically relate to your brand name.
By that I mean all possible spellings. If your name is “Markos Outils”, look at “markos outils”, “marcos outils”, “marcos-outils”, “mar cos outil”, etc.

What does this traffic mean? Today, Google is widely used as a directory to verify that a brand or reference mentioned by AI actually exists. Some users have painfully discovered the fallibility of AI when arriving in countries without a visa in 2023 because “chatgpt told me it wasn't necessary.” So today, Google is notably used to verify information transmitted by AI.

If a user wants to "verify that you exist," it's an excellent sign because it implies that they're interested enough to take the time to check. This is a crucial subtle signal for understanding where you stand today.

4. The commercial impact

As I've been telling all my clients for the past two years, the only metric (the only number) that doesn't lie is your revenue. So, if you can correlate weak signals and activity spikes with revenue or hot leads, as I explained earlier, you're onto something, and you need to dig deeper to iterate!

Above all, and most importantly, the only true truth is firsthand testimony. So ask your customers and incoming leads as often as possible: "How did you find us?"

Which pages should be analyzed first?

Contrary to popular opinion which would advise you to start with sales pages (services, etc.) because they make money, I advise you to do something else.

You are starting from scratch, given that these reports are brand new. Therefore, you must adopt a posture of observation and analysis:

which pages are coming up?
Why these pages, what type of information is highlighted on them, how is it presented?

The challenge: understanding what information actually helps Google consider your company as a credible source.

With this in mind, I will explain the (real or imagined) usefulness of llms.txt in this article.

And from there, zoom out: it means that this is the type of information that my target customers are looking for today in AI via Google AI.

Only by understanding your user, your prospect, your target customer can you design a high-performing and effective AI acquisition and visibility strategy. Otherwise, you might as well be building castles in the air.

FAQ

Where can I find the AI report in Search Console?

The report is being rolled out gradually, country by country, account by account. When available, it will appear in the performance reports of your Google Search Console, where you would otherwise see your impressions, CTR, clicks, and positions for regular Google search.

Does the report show clicks originating from AI Overviews?

No, the report only shows impressions, not clicks (especially since most users read the response and don't click anyway).

Does Search Console measure ChatGPT?

No. Search Console measures the Google ecosystem, including AI Overviews and AI mode. Other AI tools don't currently have their own tracking systems.

Should we create content specifically for AI Mode?

Google recommends prioritizing the fundamentals: accessibility, original content, expertise, consistent structured data, and useful answers to real-world questions. Specific "GEO hacks" don't replace these fundamentals. But this is just Google's perspective, not a universal one for all AI models.

Among the institutional sources

📊 What if AI and data finally became useful to your brand?

At ELLEVATE, we help you integrate these tools into your marketing, without bullshit or loss of soul, but with effectiveness.

Let’s discuss your data & AI challenges

Jade Caillot

Jade Caillot is an entrepreneur, digital growth strategist, and specialist in SEO, growth marketing, and optimization for AI and LLM. A Harvard graduate based in Toronto, at the heart of the global AI ecosystem, she develops organic growth, growth hacking, and digital strategies inspired by North American methods. She is the founder of ELLEVATE.

Google is finally measuring your visibility in its AI responses: what Search Console really shows

What does the Generative AI Performance report measure?

Why this data changes SEO reporting

What the report does not allow us to conclude