Why Did I Get a Spike in Generative AI Impressions in GSC?

The rollout of AI and its integration across the digital world has led to a lot of changes throughout marketing.

It's not just about AI content generation. We're talking about AI in analytics, AI in search, AI in reporting, and more. It's everywhere, it's inescapable, and whether or not you like it, as a site owner, you need to know what's happening.

Google recently rolled out a new report in the Google Search Console. Found in the sidebar under Search Results, you can now see a report specifically for Generative AI Search Performance.

What is this report, and what does it show you? What does it mean if you have a spike in GenAI impressions? It might not be intuitive, so let's talk about it.

What is the Generative AI Performance Report?

First, what is this report, what does it show you, and what does it not show you?

Per Google, the Generative AI Performance Report is a report showing you data on how your site is performing in Google's AI features.

It tells you:

• Your search impressions via the AI Overview and other AI features.
• Per-page impression ratings for AI features across your site.
• Basic secondary information for those impressions, like country of origin, device division, and day-by-day breakdowns.

This data comes primarily from the AI Overview box and, if users interact with it, the subsequent AI Mode. It does not cover secondary AI features like the AI-generated People Also Ask questions, nor does it cover data from Search Labs experiments.

There's also a specific, different report for AI performance in Google Discover.

One important thing to note is that this only tracks impressions and uses the same impression metrics as the normal performance report. Google doesn't differentiate between clicks through to your site that come from the AI Overview or from normal search results, so the AI Performance Report won't show you click data.

There's also one more important thing to know about the AI impressions here; they still count even if a user doesn't actually see them.

If you use Google frequently these days, you know that the AI overview generates a fairly sizable answer to most queries, but most of the answer is hidden below the "Show More" box. 

If your site is getting cited in the AI overview, but you're further down in the answer, you can be invisible until a user expands the answer. If the user doesn't expand the answer, they don't see your site. But Google doesn't care, the same way they don't care if you showed up in position #10 below the fold and the user never scrolled that far. Both will still count as an impression - similar to how high impressions don't always translate to clicks in standard search results.

FAQ on the Google Search Console Generative AI Performance Report

I wanted to answer some questions you might have, so let's go through them.

Why Doesn't My Site Have an AI Performance Report?

Right now, you should have it. Google only recently started rolling out this report, but as of the end of August 2026, it's rolled out to all websites. 

That means there are only two reasons why you wouldn't have this report.

The first is that you don't have enough impressions from AI features to be worth reporting. Small sites, sites with little or no traffic, sites in niches that don't get AI features (an increasingly small demographic of things like adult content), and new sites with no real indexation yet might all be invisible to the AI, and thus don't get the data to report.

The other reason is if you've intentionally opted out of Google's generative AI features. Google allows you to block the Google AI crawlers and, as a consequence, not appear in their AI features. Whether or not this is a good idea is a mixed bag (and I wrote about it here), but it will definitely mean you don't have data to generate a report.

How Far Back Does the AI Performance Report Go?

With a lot of analytics, you might be used to only getting data once you start gathering it, for obvious reasons. Since you already have Google Analytics and Google Search Console linked up to your site, though, you might expect to have a lot of historic data here.

In fact, you do, to an extent. Even if you only just got the report as of the end of the rollout in August, you'll have data from months prior.

There's a limit, though; Google didn't even start gathering this data globally until relatively recently. If you go to expand the timeline to see historic performance, you'll be stopped. Google will only show you data from June 2, 2025, and later. Anything before that is unavailable.

Does Google Track Other AI Impressions?

Nope. The AI Performance Report is just for Google's AI features, and not all of them at that.

Appearances in things like Perplexity, ChatGPT, Claude, or other AI avenues that can surface your website are not tracked this way in the Search Console. To see that, you need to view traffic referrer sources in Google Analytics, or set up a secondary means of tracking traffic sources.

What Does it Mean when Generative AI Impressions Spike?

So, if you're looking at your report and you see a spike in generative AI impressions, what does that mean?

Well, simply, it means the number of impressions you're getting (per day, per week, whatever dimensions you have selected) has gone up. If you're unsure how to interpret these signals, our Search Console warnings guide can help clarify what different alerts mean.

Fundamentally, that's it. 

What Causes Spikes in Generative AI Impressions?

Why might you get a spike in generative AI impressions? A few reasons. Most of them actually don't have as much to do with AI itself as you might think.

Some of your pages moved from organic results to AI results.

This is one of the big ones. Google's AI Overviews are similar to but separate from the main search results. Sites can appear in the AI overviews that aren't listed on the main search results until position 8+, or even on page 2+. Conversely, great resources ranked in the top three might not be cited by the AI at all.

One phenomenon I've seen occasionally is the move from one to the other. Sites that were once in the #1 or #2 spot drop down to #9 or #10 (or even fall off the front page entirely), but at the same time, they end up cited in the AI overview.

I suspect what's happening here is that Google is treating the AI Overview as a sort of New #1, or what we used to talk about as the Position #0 space. Since Google doesn't like to have the same domain listed more than once in the SERPs outside of special circumstances, sites that get an AI Overview listing end up dropping down organically.

This isn't guaranteed, or even super common, but I have seen it happen. At the same time, I've also seen plenty of queries where the AI Overview is basically a summary of the top three organic results, so it's hard to say what's actually happening under the hood. If you're also noticing a drop in your average position in Search Console, this could be why.

On that note, if any of you Google folks want to clarify, the comments are open!

New pages have entered the AI overviews that weren't cited before.

One of the most likely reasons why your impressions have spiked is that new pages on your site have started to be cited when they weren't before.

Note that this doesn't necessarily mean new pages, just pages that weren't cited before. This could be new pages, but I've also seen spikes happen because Google decided a three-year-old blog post happened to be great for answering a given question and popped it into the AI Overview.

You can get a decent idea of this by spot-checking before and after the spike, looking at the list of pages and their impressions. This will help you see if any one page is responsible for the spike, or if it's more of a general, across-the-board increase.

A topic is trending when it wasn't before.

Since the AI Overview impressions report is just a report on impressions, it's basically just a reflection of the traffic a query gets. If you've already been cited for queries, and the traffic for those queries spikes, your impressions will spike with it.

Think of it like a topic going viral. A lot of people are now searching for some reason, and you're getting more impressions because of it. There can be all kinds of reasons for it, from national events to holidays to seasonal shifts, to new product releases or new TV shows driving interest. All of that is basically identical to normal impression and traffic spikes.

Natural Variability

One thing to keep in mind is that the AI Overviews are not static. You can test this yourself by performing a query, and then performing the same query using an incognito window or another browser. The generated results will be different, often with entirely different cited sites.

For example, I did a search on this topic specifically, and:

• One Overview cited Edward Sturm, 1Digital, WebFX, Optimizely, and Reddit
• The other cited Pepper Content, Bradlee Bartlett, Rosemont Media, and Edward Sturm

Same exact query, different browser, ten minutes apart, and the overlap in cited sites is only one, a single YouTube video. Maybe today the AI has decided your site gets to win the jackpot. The inner workings of Google's AI are inscrutable to those of us on the outside.

How to Make Use of the Generative AI Impressions Report

So, let's say you've seen a spike in AI Overview traffic; what do you do with that information?

Well, you have a few options, ranging from cynical to potentially useful.

You can use it as a vanity metric. Some people are huge into AI, while others have more tempered expectations. If you're in a position like I am, where I'm talking to business owners about metrics and search performance, I can use spikes and growth in AI Overview impressions as a proof metric of the results I can get. Pretty much all of my clients are cited in the AI Overviews regularly these days, and the impressions tell the story.

Personally, I wouldn't trust that to be a performance metric, both because of the variability of the overviews, the fact that impressions don't mean much on their own, and the difficulty in converting AI traffic. But, if I'm talking to a CEO who is all-in on AI everything, it's a pretty good way to show that I know what I'm doing.

You can use it to see what blog formats are working the best. While the actual sites listed in the AI Overviews will change, the formatting and the data in them tend to be similar. The more informational and less sales-focused the content, and the more accessible it is, the more likely it is that the AI is going to pick it up. So, take a look at the pages being cited, and look for things like:

• How is the content formatted? Are there keys in the formatting of the information?
• Are there meta reasons, like Schema, that would make the page more attractive?
• Are there ways you present your information that are shared among top-performing posts?

When you see what the most-cited posts look like, you can then replicate that across your site and try to optimize for AI Overview citations. This will also likely optimize for other AI systems as well, since they tend to work in similar ways, even if the specifics will vary.

You can look for interesting demographic information. For example, you might have a site that is largely desktop users from the USA, but if your AI Overview impressions tend to come from mobile users in Canada, you'd have a whole different audience to consider. It could be possible that you have an untapped demographic you can identify.

Overall, though, temper your expectations. I've already talked about how variable the AI Overviews are, but it's also worth considering that impression data is not inherently meaningful in itself. You can get "impressions" for showing up on a page even if a user never actually sees your link. Is that useful to you?

On top of that, while Google is tracking impressions divided between the AI Overviews and other channels, they don't separate out click data. You have no way to tell whether your clicks are coming from organic SERPs or AI Overview citations, at least not yet. If you're trying to understand what counts as a good CTR in Search Console, that ambiguity makes it even harder to benchmark your performance.

I would guess that Google wants to add that dimension to their reports eventually, but until that day comes, I'd be cautious about using raw impression data for any meaningful decisions. To get a fuller picture of whether your efforts are paying off, it's worth stepping back and evaluating whether your blog is actually working beyond just impressions alone.