ObsurfableWhen a buyer searches your company by name, they are not browsing ten blue links anymore. In many cases, they meet Google's AI Overview first — a synthesized paragraph that frames who you are, what you sell, and which sites Google trusts to say so.
That shift accelerated sharply in September 2026. According to DemandSphere, which checks the same branded keyword set once per day across markets and devices, AI Overviews appeared on 26.12% of tracked brand-name searches on September 1. By September 27, the rate hit 90.48%, then read 80.23% and 82.06% on September 28–29. Most of the increase landed in the last two weeks of the month.
Google has not announced a change. The data measures whether an overview box appeared — not whether it named the brand, cited the official site, or got facts right. All three gaps matter for reputation and conversion.
Why branded queries are a different visibility surface
Branded search is high-intent traffic. The user already knows your name; they want pricing, reviews, locations, support, or validation before they buy.
Historically, you controlled the narrative through:
- Your homepage and key landing pages
- Your Google Business Profile
- Review platforms and directories that rank for
[brand] reviews
An AI Overview sits above that stack. It is editorial copy Google generates from whatever sources its systems retrieve — which may include competitors, aggregators, old press, or third-party listicles. You do not approve the wording.
That is not hypothetical anymore for most brands DemandSphere tracks. If the September curve holds or spreads, "search our name" becomes "read Google's summary of us" as the default first step.
What DemandSphere measured (and what it did not)
DemandSphere's series is useful because it is longitudinal on a fixed branded keyword set, with one check per keyword per day. Key properties:
| Property | Detail |
|---|---|
| Query type | Brand-name searches in their tracked portfolio |
| Frequency | Once per day per keyword |
| Coverage | All markets and devices in their methodology |
| Trigger | Any AI Overview on the SERP counts — with or without a brand citation |
| Google signal | No public announcement tied to the jump |
What the dataset does not tell you:
- Whether the overview mentions your brand by name
- Whether it recommends you vs a competitor
- Accuracy of claims (pricing, features, leadership, locations)
- Click-through impact on branded sessions
Those require your own audit loop — the same way you would not trust a rank tracker to substitute for reading the SERP.
Agency AEO's October 5, 2026 brief notes complementary spot checks: search consultant Chris Long reportedly found AI Overviews on 93 of 100 brand queries in a separate sample. Directionally, multiple observers saw the same pattern in late September.
How this connects to broader AI Overview behavior
The branded surge sits on top of a year of expansion:
- I/O 2026 merged AI Overviews with AI Mode into one integrated AI search experience for 2.5B+ monthly users (Keywords Everywhere tracker, updated October 2026).
- August 2026 product changes expanded full AI-style answers without an extra "Show more" click on some queries — pushing organic results down (October 2026 Google search roundup).
- Independent measurement still finds question-form queries trigger overviews far more often than the overall average — 64.7% vs 13.7% in the WashU / arXiv audit of 55,393 trending US queries (March–April 2026).
Branded queries were never the highest-AIO category in that academic sample — trending news and informational mixes dominated. September's tracker data suggests Google widened the trigger set toward navigational and brand-intent SERPs without publishing new guidance.
For brands, the practical merge is simple: you already lived in AI Overviews for "how does X work" questions; you may now live there for "what is X" too.
What can go wrong on a branded SERP
When the overview is wrong, you cannot edit it directly. Failure modes that show up in brand audits:
- Stale positioning — acquired product lines, old pricing, deprecated features
- Wrong entity — name collision with another company
- Competitor-framed comparisons — "alternatives to X" logic imported into a branded query
- Missing differentiation — generic category description that fits any vendor
- Ghost attribution — your facts appear without your brand named (ghost citations)
Google's overview may cite credible domains while still mis-synthesizing details. WashU researchers found 11.0% of atomic AI Overview claims unsupported by cited pages in their crawlable sample — with a stated lower bound around 5.3% after accounting for excluded social/video hosts (arXiv:2605.14021). Branded summaries are not exempt from synthesis error.
A weekly branded AI visibility panel
Treat branded AI Overviews like a new owned channel you influence indirectly. A minimal operating panel:
Queries to run (private window, logged out + logged in)
[brand][brand] reviews[brand] pricing/[brand] cost[brand] vs [main competitor][brand] near me(if local)[brand] support/[brand] login
Record for each: overview present? brand named? recommended? top three cited domains? factual errors?
Sources to fix first
When the overview is wrong, trace citations backward:
- Your site — About, pricing, product, FAQ, schema, dates
- Google Business Profile — categories, services, hours, description
- High-citation third parties — G2, Capterra, Wikipedia, Reddit threads, news profiles
- Aggregator listings — stale duplicates outrank your canonical facts
Use the SERP feedback control on the overview when claims are clearly false — then re-check in 7–14 days. AI summaries drift.
Metrics in Search Console
Watch branded query impressions, CTR, and average position alongside the manual panel. A branded overview can lift impressions while suppressing clicks — the Pew zero-click pattern is not brand-safe just because the query includes your name.
Separate Google AI performance reporting (rolling out globally in fall 2026 per industry trackers) from ChatGPT / Perplexity mention rate. Branded Google search is only one engine's front door.
Playbook by team
Marketing and comms
- Add branded AI summary accuracy to quarterly brand health reviews
- Align press releases and About-page updates with the exact phrases buyers search
- Prioritize third-party corroboration on claims you need Google to repeat (awards, customer count, compliance)
SEO and content
- Ensure crawlable, dated, canonical facts on pricing and product boundaries
- Publish clear disambiguation if you share a name with another entity
- Avoid self-ranking "best of" listicles as your primary reputation strategy — September 2026 practitioner research found AI Overviews citing those lists less often and omitting self-ranked publishers more frequently (Lily Ray analysis, 100 B2B software queries)
Revenue and CX
- Train support to ask "what did you read about us before you called?"
- Map common overview errors to macro responses and site fixes
How Obsurfable fits
Branded Google search is one surface. Buyers also ask ChatGPT, Perplexity, Gemini, and Claude who to trust before they ever type your domain. Explorer stores public observations — prompts, answers, brands mentioned, citations — so you can compare what Google summarizes with what conversational engines recommend.
The App matches that corpus to your company for mention rate, share of voice, and source-to-mention paths. When September's branded-overview expansion changes first impressions on Google, you still need repeated prompt runs elsewhere to see the full picture.
FAQ
Did Google confirm the branded AI Overview increase?
Not as of October 2026. The shift is documented by third-party rank tracking (DemandSphere) and spot audits, not a Search Central post.
Should we panic if there is no overview on our brand today?
No. Rates fluctuate by query, locale, and account. Establish a baseline panel and watch week-over-week presence — not a single screenshot.
Is fixing our website enough?
Often necessary, rarely sufficient. Overviews draw on third-party evidence. Combine owned-site accuracy with reviews, press, and directory consistency — the same footprint that drives earned citations.
Does this replace GEO work on ChatGPT?
No. Engines disagree on who they recommend. Branded Google monitoring is table stakes, not the whole program.
Bottom line
In September 2026, AI Overviews went from ~26% to ~80–90% of DemandSphere's tracked brand-name searches — without a Google announcement. Presence is not accuracy, and citations are not control. Search your name the way a customer would, fix the sources Google reads, and measure conversational AI recommendations on the same buyer questions. Your first impression is increasingly generated, not designed.