AI Overviews on Brand Queries: 83% Presence, and What About CTR?

What Does the Ahrefs Study Really Measure on Brand Queries?

A client calls me on a Tuesday morning. He opens his Search Console in front of me. Impressions on his brand name hold steady. Clicks to his site are sliding. He asks me what broke.

Nothing broke. Something landed on top of him.

On October 2, 2026, Ryan Law publishes a study on the Ahrefs blog that puts a number on what my client was seeing: AI Overviews appear on 82.91% of Google US desktop results for brand keywords, as of September 29, 2026. 83%. That number marks a change of state. Not just one more statistic.

It all starts with a test by Chris Long. Out of 100 company brand names, 93 triggered an AI Overview. Ahrefs wanted to know whether that test held at scale.

It holds.

Here is the sample. 232.8 million Google US desktop result crawls, from July to September 2026. Within it: 92.5 million crawls on brand keywords and 140.2 million on non-brand keywords. The volume is solid. The scope is clear: United States, desktop, one quarter. A brand query, here, is the name alone (« bluehost ») but also any search that contains the brand.

Why focus on brand queries? Because your name is the most profitable query you own. Nobody bought it per click. It comes from everything else: word of mouth, advertising, press, the product itself. A visitor who types your name arrives with the strongest intent that exists. They aren’t looking for an answer. They’re looking for you.

Until now, Google let them find you. Your site showed up first, the visit followed. Today, a generated summary steps in between. It answers before the click. And it cites whoever it chooses.

The problem isn’t your ranking. It’s what displays before it.

This article walks through the Ahrefs data, what it says about CTR, and what you gain by doing this week. One caveat first: the study measures the presence of AI Overviews, not the clicks lost. You won’t find any loss figure here that nobody has published. The figures cited come from Ahrefs. The rest is field observation, flagged as such.

En trois mois, l’écart s’effondre. Les AI Overviews passent de 61,3% à 73,3% sur marques, et la différence avec non-marques tombe de 12,2 points à 5 points. Votre requête marque n’est plus un refuge.

From 61.3% to 73.3% in Three Months: The Gap with Non-Brand Is Closing

Look at the trajectory. Not the isolated number.

Brand: 61.3% in July. 65.1% in August. 73.3% in September. That’s 12 points gained in three months, roughly 20% relative growth over the July rate, according to Ahrefs. Non-brand: +4.8 points over the same period.

The key signal is elsewhere. In the gap between the two segments. 12.2 points in July. 9.6 points in August (calculated from the published rates). 5 points in September.

« The gap between the two groups went from 12.2 points in July to just 5 points in September. » — Ahrefs, free translation

In its previous research on AI Overview triggers, in November 2025, Ahrefs found AI Overviews far more frequent on non-brand queries than on branded ones. That is clearly no longer the case, writes Ryan Law. With an honest caveat: the samples differ from one study to the next. The comparison gives a direction, not a measurement to the tenth of a point.

My reading, and it’s an opinion: Google long treated the brand query as a navigational query. A name typed, a site expected. It now treats it as a question. Who is this brand? What does it sell? What do people say about it? A question calls for a summary. And a summary displays above the links.

« Informational versus navigational » is no longer the diagnosis. What matters is the surface occupied by the generated answer, whatever the intent.

For years, the brand has been SEO’s last safe harbor. Everything could move on generic queries; the brand name brought its click home. Today that harbor shows an open door in three months of data.

Three months is short. Ahrefs also points out that Google may decide to move these blocks lower on the page. A three-month slope is not a law. But waiting for the next quarter to find out means choosing to discover the effect in a year-end report. You gain by measuring now, while your comparison history is still clean.

One simple question for your next report: do you still separate brand traffic from non-brand traffic, month by month? If the answer is no, the most informative line in your dashboard is missing.

Sur les 100 marques du panel, 76% des AI Overviews occupent la position 1 – la place avant votre résultat traditionnel. Une concentration qui accentue l’impact sur le CTR.

Position 1: Why 48 AI Overviews Out of 63 Change How You Read Your SERP

An AI Overview can appear in several places on a results page. They are not all equal.

Ahrefs says it clearly: its research suggests that AI Overviews reduce click-through rate. The worst case is an AI Overview in position 1 on your brand query. In place of your domain.

To measure this case, Ahrefs put together a second sample. 100 brands drawn from the Interbrand Best Global Brands 2025 ranking. Bare-name queries, like « bluehost », checked at 14 weekly checkpoints between July and September.

From 5 to 63 in one quarter. The appearance accelerated in early September. The rise was fast, not gradual.

48 out of 63. Position 1. Average 1.49.

What does the user see then? A block of generated text, half a screen tall on desktop. Below it, sometimes, your link. When I look at these pages on my clients’ brands, the first reaction is always the same: « but we’re still first. » Yes: first in the ranking, but not the first element the eye lands on.

That’s the shift. Being first in the organic results no longer means being the first element read. The ranking hasn’t changed. Only the visual hierarchy has.

Two limits to keep in view. First, this panel is made up of very large global brands: a local SMB won’t necessarily recognize its own case in it. Second, it covers bare names, not compound queries like « brand + reviews » or « brand + price ». Ahrefs flags this itself.

But these numbers are worth your attention. When 63 brands out of 100 trigger the block, and three times out of four it takes the top of the page, your brand will be part of it. When? Through which sources?

And that point leads us to the real question: when a summary speaks on your behalf, who supplies its material?

Wikipedia, YouTube, LinkedIn: Who Speaks in Place of Your Brand?

This is the risk I find most underestimated. Not the lost click. The borrowed voice.

In the same 100-brand panel, Ahrefs looked at external citations, meaning links to sites other than the queried brand’s own. The three most frequent sources: Wikipedia, YouTube and LinkedIn.

Ahrefs frames the risk this way: directing the user to unofficial sources or, in the worst case, to competing brands. Wikipedia, for its part, is a fairly neutral source. But neutral doesn’t mean up to date. Or complete. Or aligned with your current positioning.

When I test a client’s brand name in Google, I no longer look only at where their site ranks. I read the generated block. I note the cited sources. I compare what it says with what the client would want said. The gap is often there, and it’s invisible in any rank-tracking tool.

Why these three? Because they give a clear, current description. An AI finds what it needs there without crossing ten commercial pages. Your site, meanwhile, often speaks the language of sales. Not the language of definition.

The diagnosis is simple: the problem isn’t that Wikipedia gets cited. It’s that your own site isn’t the easiest source to cite.

And this can be fixed. An « about » page that says who you are in two sentences. One page per major brand question: price, reviews, alternatives, delivery, contact. Consistent vocabulary from page to page. These are semantic cluster building blocks applied to your own name, and they give Google what it needs to cite you, rather than a third party.

When you type your brand today, do you know which sources feed the answer? If you haven’t checked since September, check before the end of the week.

What Is the Real Impact on CTR? What We Know, What We Don’t

Let’s come back to this article’s title. « What is the real impact on CTR? » The honest answer: the study doesn’t quantify it.

Ahrefs measures the presence of AI Overviews. Ahrefs notes, based on its earlier research, that they reduce click-through rates in général. But the percentage of clicks lost on brand queries, this study doesn’t publish. I won’t invent it either.

What we can understand, however, is the mechanism.

A brand query carries three kinds of intent. Reaching the site (the name alone). Learning about the brand (« brand reviews », « what is brand »). Comparing or buying (« brand price », « brand vs competitor »). The generated block helps most for the last two. It summarizes, it compares, it answers. The user no longer has to click to get the essentials.

I observe a trend among my clients that supports this analysis: compound brand queries are the first where the gap between impressions and clicks widens. Field observation, not a published statistic. It’s up to you to confirm or refute it on your own site, which is precisely the point of the method below.

One detail changes the reading. A falling CTR doesn’t mean falling demand. Brand impressions reflect awareness. If they hold while clicks decline, your brand hasn’t lost audience. The visit happens elsewhere, or later, or in another form: a phone call, a direct search, a detour through a third-party source.

The trap would be to treat the decline as a classic SEO problem. Redoing titles. Rewriting meta descriptions. Pushing content. None of that moves a generated block that steps in before your link.

Your click-through rate isn’t the problem. The problem is that the answer displays before your link.

There is good news too, and I take it seriously. A generated block cites sources. Your domain can be one of them. A link in the block beats no presence at all, and you gain by structuring your pages to earn that citation. For brands that anticipate, this is a visibility opportunity, not just erosion.

So, how much of your brand traffic still depends on a click on the first blue link?

Five Actions to Take Back Control of Your Brand Name

Enough theory. Do this, in this order.

1. Measure in two groups. Name alone on one side, name + modifier on the other. This split shows you where the erosion comes from: navigation or information. One hour in Search Console. Months of guesswork saved.

2. Take a weekly snapshot. Ahrefs tracked 14 checkpoints across 100 brands, with weekly readings. At your scale, it’s simple: capture the SERP for your name every Monday, dated and archived. Three months later, you’ll know which block appears, where, and from which sources. Data your competitors don’t have.

3. Audit the cited sources. Wikipedia, YouTube, LinkedIn, as we saw. Fix the errors, fill the gaps. On each platform, patience wins. Forcing it gets you nowhere.

4. Build your brand cluster. I build clusters for product catalogs. Apply the same logic to your name: a central page that says who you are, then satellite pages for each question tied to your brand. Price, reviews, alternatives, delivery, warranties, contact. On each one, a crisp answer at the top, then the detail. A generative engine easily finds what is well organized.

5. Change your barometer. Measuring only organic brand traffic is looking through a filter. Also track calls, quote requests, direct searches, mentions. You then track the real demand, not just the clicks.

A word on mindset. That 82.91% figure is worrying. But above all it brings clarity. While most sites will see the effect at year-end, you’ll already have a few weeks’ head start. This is the moment when fixing costs the least.

This is not a sales pitch. I show you the pages. In a live audit, I open your brand’s SERP with you, read the block line by line, and spot who speaks on your behalf.

Tonight, when you type your name into Google, who speaks first: your site or a summary written from other pages?

Stéphane Jambu

SEO & AI Engineer

I build growth systems / AI / Neuroscience | 650+ clients · 80 LinkedIn testimonials · 30 years of expertise · 15 years of systems running without me.

Follow on LinkedIn