Google now puts an AI Overview on your brand name. Who writes it, and what you can change

KEY TAKEAWAYS

1. AI Overviews reached about 82% of branded searches by 29 September 2026 in two independent datasets. Google has not announced a change.
2. The click effect on branded searches is unmeasured. Compare your own branded click-through rate from before 18 September with the days after 27 September.
3. Third parties write most of the answer. Wikipedia appeared in 71.9% of branded source lists, YouTube in 38.6% and LinkedIn in 22.8%.
4. The Search Console exclude control removes your site as a source, not the overview. Fix the source of a wrong claim instead.

Until this autumn, a search for a bare brand name rarely carried an AI Overview. That changed in the last week of September 2026. Most of what has been written since repeats one chart. This article reconciles the two datasets behind it, separates what is measured from what is assumed, and gives a method for the part you can change, which is the set of sources the answer is built from.

What changed in September 2026?

DemandSphere tracks a set of branded keywords once a day across all markets and devices. On 1 September, 26.12% of them returned an AI Overview. The share stayed between about 26% and 35% until the 17th, spiked above 55% on three days, then stepped up. It reached 69.21% on 26 September, 90.48% on the 27th and 82.06% on the 29th. By our arithmetic the last reading is 3.1 times the first.

Chris Long of Nectiv flagged it publicly. Search Engine Roundtable quoted his post on 30 September: "Google is starting to show AI Overviews for a ton of branded queries." He ran 100 enterprise brand names through a search results API and found an overview on 93.

Ahrefs published a larger dataset on 2 October. Across 232.8 million US desktop crawls from July to September, 92.5 million of them for branded keywords, AI Overviews appeared on 82.91% of branded results on 29 September.

Search Engine Land reported that Google has not announced any change. A month earlier Google did confirm that some overviews now open at full length by default, which pushes the ordinary results further down.

Why do the trackers disagree about where it started?

The two datasets meet at about 82%. They start in very different places.

Three things explain most of the gap.

Neither firm defines a branded keyword. A set heavy with "brand plus pricing" and "brand plus reviews" behaves like any question. A set of bare names does not. Ahrefs's own figures show this. Its broad branded set was already at 61.3% in July, while its bare-name panel stood at 5 of 98, which is about 5%.

The markets differ. Ahrefs measured US desktop only. DemandSphere blends every market and device, and its breakdowns are still to come.

The brand lists differ. Long found 93 of 100 and Ahrefs found 63 of 100 for bare names at the end of September, using different lists and tools.

What the evidence supports is this. Brand-plus-question searches in the US were mostly covered before September. The new part is the bare name, and probably more markets. We cannot prove the second half from outside. A year earlier, Ahrefs found overviews on 13.1% of branded results and 0.9% of navigational ones in 146 million September 2025 searches, so the direction is not in doubt.

Position is also disputed. In the Ahrefs panel, 48 of the 63 overviews held position one. Barry Schwartz saw Adobe's at the top and most others mid-page or lower. Check your own name on a phone and a desktop.

Will this cost you branded clicks?

Nobody knows yet. Neither DemandSphere nor Long measured clicks, and three older studies point in different directions.

• Ahrefs, February 2026. Across 300,000 keywords, the top result's click-through rate was an estimated 58% lower where an overview appeared. The sample was informational keywords only.
• Amsive, April 2025. Across roughly 700,000 keywords on 10 sites, only 4.79% of branded keywords triggered an overview, and those saw click-through rate rise 18.68%. That describes a time when an overview on a brand search was rare and probably selective.
• Seer Interactive, April 2026. Across 53 brands, informational searches where the brand was cited in the overview had a 2.07% organic click-through rate, against 0.94% where it was not and 3.35% with no overview. Seer warns that one account makes up 47% of the uncited segment, and it does not break out branded searches.

None of these measures a navigational search with an overview on it. Ahrefs itself argues that the bigger concern on branded searches is how third-party sources present the brand. That is a reasonable view, not a click measurement. Measure your own.

Who writes the answer about your brand?

Mostly not you. Ahrefs looked at the 57 panel overviews that listed sources.

Competitors are scarce. Only one competitor-owned site appeared across the 57 lists, and 2 of 58 answers named a rival. Ahrefs does not report how often the brand's own site was cited. These are household names. A smaller company will have a different mix, and the only way to know it is to look.

On accuracy, the best public evidence is a study Seer Interactive ran on itself. It collected 28,123 responses to 1,562 prompts about its own company across six AI surfaces from 1 May to 9 June 2026. When the engines answered, they were right 95.9% of the time on average. AI Overviews had the highest error share of the six at 5.7%, against 1.7% for AI Mode and 1.3% for ChatGPT.

The errors were ordinary. A former office was given as current in 223 responses. Services the firm does not sell appeared in 109. A claim about client turnover traced to one review from 2018. This is one brand studying itself, so treat the rates as an illustration.

It is the same mechanism we described in how AI engines choose which brands to name, applied to your own name. It is also why brand entity and search presence are separate pillars of digital authority. Page one for your name now includes a paragraph somebody else sourced.

Can you switch it off?

Not for your brand name. There is a control, and it does something narrower.

Google announced a Search generative AI control in Search Console on 3 June 2026, first for a subset of UK site owners after a requirement from the UK Competition and Markets Authority. Search Engine Journal reported it on accounts outside the UK from July. It has three settings: include, which is the default, exclude, and inherit from a parent property. Exclude removes your site's links and content from AI Overviews, AI Mode and the generative features in Discover. Google says the setting is not a ranking signal elsewhere in Search. It works per property. Page-level controls are not due until March 2027 under the regulator's timeline. It is separate from Google-Extended, which governs training.

The control decides whether your pages can be a source. Nothing in the reporting says it decides whether an overview appears for a query. Our inference is that excluding a site leaves the overview on its brand name in place and leaves it with third-party sources only.

What can Search Console show you?

Less than you would hope.

The branded queries filter, open to all eligible sites since March 2026, splits clicks, impressions and click-through rate into branded and non-branded. Google classifies queries with an AI-assisted system and says some may be misclassified. Low-volume properties may not have it.

The generative AI report reached all sites on 31 August 2026. It shows impressions by page, country, date and device. It has no clicks, and coverage of its launch found no query data. John Mueller confirmed on Reddit in September that an overview is tracked as one block, so every link in it gets the same position.

So Search Console cannot say which brand searches show an overview. It can show whether branded click-through rate moved when the overviews arrived.

A seven-step branded answer audit

1. List 15 to 25 branded searches. The bare name, common misspellings, and the name plus pricing, reviews, login, alternatives, "vs" your nearest rival, and "is it legit".
2. Run each one signed out, on mobile and desktop. Record whether an overview appears, where it sits, its full text and every source. Repeat on three days, because presence moved by 20 points or more between consecutive days in September.
3. Break the text into claims. Mark each one correct, outdated, wrong or unflattering but true.
4. Trace each claim to a source and classify it using the table below.
5. Fix in order of speed. Your own pages first, then profiles and listings you control, then third parties.
6. Check clicks. Using the branded filter, compare click-through rate for 1 to 17 September with the days from 27 September on. Split bare-name from modifier searches.
7. Repeat monthly. The same sampling rules apply as in measuring AI visibility.

For Wikipedia, the fix is usually a better independent source, not a rewrite by your own team.

A worked example with invented numbers

This company is hypothetical. Before the change its bare name drew 8,000 impressions a month at a 55% click-through rate, which is 4,400 clicks. Its brand-plus-modifier searches drew 4,000 impressions at 30%, which is 1,200 clicks.

After 27 September the bare name falls to 52%, or 4,160 clicks. The modifier searches fall to 24%, or 960 clicks. Each group lost 240 clicks. The total moved from 5,600 to 5,120, a fall of 8.6%.

The two losses mean different things. On the bare name, 3 points went missing and the site link was still on the page. On the modifier searches a fifth of the clicks went, and those are the searches where the overview answers the question. If its pricing claim is two years old, the lost click matters less than the wrong number the searcher left with.

Where this stops

Google may pull this back. Ahrefs notes that it changes overview triggers often. The daily series was volatile before the step, and nobody outside Google knows whether 82% will hold.

The audit is worth doing either way. The sources that feed your brand's overview are the same ones other AI engines read. If you want that checked on a schedule, the waitlist is open.

Questions

How many branded searches show an AI Overview now?

DemandSphere recorded 82.06% of its tracked branded keywords on 29 September 2026, up from 26.12% on 1 September. Ahrefs recorded 82.91% of US desktop branded results on the same day. For bare brand names, Ahrefs found an overview for 63 of 100 global brands on 30 September, and Chris Long of Nectiv found 93 of 100.

Do AI Overviews on brand searches reduce clicks to my site?

Nobody has measured it for this change. Ahrefs's 58% figure covers informational keywords only. An Amsive study from 2025 found branded click-through rate rose 18.68% where an overview appeared, at a time when only 4.79% of branded keywords had one. Compare your own branded click-through rate before and after 26 September.

Can I remove the AI Overview from searches for my brand?

No setting does that. Search Console has a property-level control that excludes your own site from AI Overviews and AI Mode. It removes your pages as a source. It does not stop Google answering from other sites, and Ahrefs found Wikipedia in 71.9% of the branded source lists it examined.

What should I do if the AI Overview says something wrong about my company?

Find the source first. Seer Interactive sorts errors into three causes, which are a misread of your own page, a stale third-party source and old training data. The first two can be corrected in days to weeks by fixing the page or listing the answer draws on. The third takes months.

Sources

1. DemandSphere, AI Overviews on branded queries tripled in September demandsphere.com
2. Ahrefs, AI Overviews Currently Appear for 83% of Branded Searches ahrefs.com
3. Ahrefs, AI Overview triggers ahrefs.com
4. Ahrefs, Update: AI Overviews Reduce Clicks by 58% ahrefs.com
5. Search Engine Land, Google AI Overviews jump from 26% to 80% of branded queries searchengineland.com
6. Search Engine Roundtable, Google AI Overviews Showing For Many Large Brand Names seroundtable.com
7. Amsive, Google AI Overviews: New Research Reveals How to Navigate Click Drop-Off amsive.com
8. Seer Interactive, AIO Impact on Google CTR: 2026 Update seerinteractive.com
9. Seer Interactive, Study: AI Declined to Answer Prompts About Our Brand 31% of the Time seerinteractive.com
10. Search Engine Journal, What Opting Out Of Google's AI Search Features Means Now searchenginejournal.com
11. Search Engine Journal, Google Search Console AI Reports Rolled Out Worldwide searchenginejournal.com
12. TechWyse, Google Launches Dedicated Generative AI Search Performance Reports in Search Console, Tests Site Opt-Out Toggle techwyse.com
13. Herdl, Search generative AI controls rolling out beyond the UK herdl.com
14. Relevant Audience, Mueller: Search Console gives every AI Overview link the same position relevantaudience.com
15. Search Engine Land, Google Search Console branded queries filter expands searchengineland.com
16. Search Engine Land, Google is dynamically expanding AI Overviews for some queries searchengineland.com

NEXT STEP

See where your brand stands.

Your Digital Authority Score, who AI recommends for your target questions, and the placements that close each gap. Your first report is free.

Get your free report →