Build1 publisher2 min readPublished
AI Overviews more than tripled on branded Google searches in September, DemandSphere's tracking shows
DemandSphere's daily tracking found Google AI Overviews on 82.06% of its branded keywords on September 29, up from 26.12% on September 1. The count records only whether an Overview appeared, so each brand has to measure the effect on its own clicks and listings.
The Engineer · Build desk
Drafted by a language model from the sources cited here and checked against its claim ledger before publication. How we use AISend a correction
What happened
- Presence on DemandSphere's tracked branded keywords peaked at 90.48% on September 27.
- DemandSphere counted an Overview whenever one appeared for a tracked query, whether or not the searched brand was named in the generated answer.
- The readings are daily, per-keyword measurements from DemandMetrics, pooled across all markets and devices.
- Google had made no official announcement of the change when DemandSphere published its analysis on October 1.
Compiled by The EngineerSomething wrong?How this is made
Why it matters
- decision One manual look at a brand's results page is a weak sample when the pooled share fell 8.42 points in the two days after its peak, so checks have to repeat across days.
- exposure A company's own name can bring up an AI-written answer above its organic listing that does not name it, link to it, or describe it correctly.
- constraint The series cannot size any traffic loss. That estimate needs a brand's own impressions, clicks and landing-page data, because the series records no clicks or citations.
On the series' own endpoints, DemandSphere's description of AI Overviews having "more than tripled" for branded queries holds [4]. Divide the September 29 reading by the September 1 reading and you get about 3.14 [1]. Against the September 27 peak, the ratio is about 3.46 [2]. In percentage points, presence rose 55.94 over the month [3].
The peak did not hold. According to the post, the lower readings on September 28 and 29 suggest presence can fluctuate even after a major expansion [10]. It also says the increase was rapid, not gradual [17].
Treat the 82% as a claim about DemandSphere's keyword set [3]. For it to describe one brand, that brand's queries would have to look like the tracked set. Its traffic would also have to look like a mix pooled across every market and device. The write-up is plain about its limits. The post does not say how many keywords the set contains [16]. It says the published data do not establish whether the pattern applies evenly across locations, devices, industries or individual brands [15], and there is no immediate country-level or device-level split [9].
The data also cannot give a cause. Daily tracking records what Google served on a given day. It cannot explain the internal Google changes behind the pattern [18].
The post lists four checks [13]: whether Overviews appear for the company name, product names and brand-plus-service searches; whether the Overview describes the business accurately; whether the brand's own pages are cited or third-party sources dominate the answer; and whether changes in branded impressions, clicks and landing-page traffic coincide with the rise. The first check is DemandSphere's own yes-or-no reading, run on a keyword list of one. The middle two are about what the Overview actually says. In my view the fourth matters most. It is the only check that tests for what the post calls an actual performance problem, as distinct from a visible interface change [14].
What to watch
- Whether Google publishes any statement or documentation covering AI Overviews on branded queries.
- DemandSphere's October readings, and whether presence stays near 82% or swings back toward the start-of-September level.
- Any country, device or click-level breakdown that would show whether the rise changed traffic to brand sites.