Skip to main content
Answer Engineered

ENTRY 018 · GEO · By Answer Engineered Research

2vs1

An SEC Filing Credits McKinsey for a 60% Stat. The Number Is Bain's.

CREDITED TO MCKINSEY 2FOUND IN THE MCKINSEY ARTICLE 1

Change Agents Corporation's S-1/A sells a GEO product on two statistics credited to McKinsey. The report it links contains one. Bain published the other.

· 10 MIN

Where the sentence sits in the filing

It is not a footnote in an appendix. It is the opening of the section that describes the product, headed “The Beacon Agentic Generative Engine Optimization (GEO) Search Product”, and the same passage appears twice in the document — once in the Prospectus Summary near the front, once again in the Business section.

Both times it does the same job. The two numbers establish that a problem exists, the next sentence says small businesses cannot see it — “Most small businesses are invisible to these systems and have no idea that this problem even exists” — and the product follows immediately:

“The Beacon Agentic GEO search product is designed to help small, local service businesses get recommended by AI systems like ChatGPT, Gemini, Claude, Perplexity and Grok. The agentic software product is designed to autonomously; a) diagnose a small businesses AI visibility and SEO effectiveness in about 60 seconds, b) provide a visibility score, and c) generate implementation-ready fixes upon approval.”

The filing adds that “The Beacon app will be offered in the form of a monthly subscription fee.”

This company has been in this blog before. Its Form 10-Q, filed 14 August, was the first we found to name generative engine optimization as a business segment rather than mention it in passing. The 10-Q described the product. The S-1/A is where the market case for it gets written down.

The 50% is McKinsey’s, and it holds up

Start with the number that checks out, because the contrast is the point.

McKinsey’s article is “New front door to the internet: Winning in the age of AI search”, dated 16 October 2025. Its opening statistic card reads “50%” / “of consumers” / “already use AI-powered search today (Source: McKinsey AI Discovery Survey, n = 1,927)”. The body says: “Half of consumers polled in a McKinsey survey now intentionally seek out AI-powered search engines, with a majority of users saying it’s the top digital source they use to make buying decisions.” The footnote attached to that sentence reads: “McKinsey’s AI Discovery Survey was fielded among a representative US consumer panel during August 2025, n = 1,927.”

Later in the same piece there is a second line the filing’s phrasing could rest on: “around 40 to 55 percent of consumers in top sectors (consumer electronics, grocery, travel, wellness, apparel, beauty, and financial services) are using AI-based search to make purchasing decisions.”

So the filing’s first sentence is a fair paraphrase of a real, sourced McKinsey finding. We are not going to pretend otherwise.

What separates the two documents is not accuracy. It is disclosure. Every statistic in the McKinsey article carries a numbered footnote — ten of them — naming what produced it: a survey with a panel and an n, a McKinsey projection, a McKinsey analysis, a CMO survey of “Fortune 500 consumer brand CMOs in September 2025, n = ~30”. The filing carries two statistics and one orphan URL.

One more thing the article does contain, since the filing was reaching for a market-shift figure: “unprepared brands may experience a decline in traffic from the traditional search channels: anywhere from 20 to 50 percent”, footnoted “McKinsey analysis”. There was a McKinsey number available about traffic loss. It is in the document the filing linked.

The 60% is Bain’s

Bain & Company published a press release on 19 February 2025 headed Consumer reliance on AI search results signals new era of marketing. Its second bullet: “On traditional search engines, about 60% of searches now end without the user progressing to another destination.” Its body: “Bain finds 60% of searches now terminate without the users clicking through to another website.”

Set that next to the filing: “60% of searches now end without a click”.

Bain: “60% of searches now end without the user progressing to another destination.”

Six words survive intact, in order: 60%, of, searches, now, end, without. That is not proof of a path from one document to the other, and we are not claiming one. It is why the number was findable at all.

Two differences are worth marking. Bain writes “about 60%”; the filing writes 60%. And Bain scopes the claim to traditional search engines; the filing scopes it to searches.

Now the part that matters more than the attribution. The Bain release discloses no methodology whatsoever. No sample size. No field dates. No countries. No definition of a search or of a destination. It reports that consumers “rely on AI-written results for at least 40% of their searches, reducing organic web traffic by 15% to 25%”, and it points to a longer Bain research brief, which we did not open and are not citing here. The release is the form in which this number travels, and in that form it is unauditable.

So the filing swapped a footnoted survey figure for an unfootnoted one, and credited both to the firm that published the footnoted one.

The two statistics, side by side

Statistic as the filing states itFirm that published itMethodology disclosed at the sourceDate
”50% of consumers use AI-powered search to make buying decisions”McKinsey & CompanyAI Discovery Survey, representative US consumer panel, August 2025, n = 1,92716 October 2025
”60% of searches now end without a click”Bain & CompanyNone in the release: no sample size, no field dates, no country scope19 February 2025

And the counts behind that table, each one a plain search of the two documents:

Counted in the filing and in the article it linksCount
Statistics credited to McKinsey in the passage2
Of those, statistics found in the McKinsey article1
Numbered footnotes in the McKinsey article10
Occurrences of “Bain” in the filing0
Occurrences of the numeral 60 on the McKinsey page0

The citation for both statistics is a single line of text sitting alone in its own paragraph, immediately after the Beacon description:

https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/new-front-door-to-the-internet-winning-in-the-age-of-ai-search?

stcr=6CA8D88A0086426FAC786157DD60E8C1&cid=mgp_opr-eml-alt-gms-mgp-glb--&hlkid=cf3225700e534c3e94beae40d942e868&hdpid=a892f2ad-facb-40fc-9a55-474378d99663

Four observable properties of that string, all checkable in the filing’s HTML.

It is not a hyperlink. The document contains no anchor element pointing at mckinsey.com anywhere. A reader of the prospectus cannot click it.

It is split by a literal line break, inserted immediately after the question mark, in the middle of the query string. The break above is where the filing puts it.

It carries four parameters: stcr, cid with a value containing the token eml, hlkid and hdpid. A URL does not acquire campaign parameters by being typed. It acquires them by being copied from somewhere that added them.

And it carries no citation apparatus at all: no “Source:” label, no footnote marker, no article title, no author, no date. A reader who wants to know which of the two statistics this URL is supposed to support has to open it and find out.

What that establishes is narrow: the URL was copied rather than composed, and it was never checked against the sentences above it. What it does not establish is who wrote the paragraph, what they read, or what they meant to do. We have no evidence on any of that and are not offering an opinion.

Who is making the pitch

All of the following is disclosed by the company itself, later in the same document.

The filer was called Avalon GloboCare Corp. until 20 July 2026, and its ticker changed from ALBT to CHGA on 22 July 2026. On the cover page: “The last reported sale price of our Common Stock on Nasdaq on August 24, 2026, was $0.17 per share.”

In the risk factors: “On April 15, 2026, we received written notice from Nasdaq that we were not in compliance with Nasdaq Listing Rule 5550(a)(2), as the minimum bid price of our common stock had been below $1.00 per share for 30 consecutive business days.” The filing states the company has until 12 October 2026 to regain compliance. That is a deficiency notice with an open cure period. It is not a delisting, nothing in the filing says one has been decided, and neither do we.

Also in the risk factors: “There is substantial doubt about our ability to continue as a going concern”, alongside “net losses from continuing operations of approximately $4.77 million and approximately $5.69 million for the six months ended June 30, 2026 and 2025, respectively.”

None of that makes the citation wrong. The citation is wrong on its own evidence. What the context answers is a different question: how much scrutiny a market-sizing sentence gets when it moves from a landing page into a registration statement. Here it went in twice, unchecked, in a document reviewed before filing.

What this evidence cannot tell you

We did not test Beacon. No claim about the product is evaluated here — not the 60-second diagnostic, not the visibility score, not the fixes. Those are the company’s own descriptions of a product it says is still in development, and we have no independent data on any of them. A wrong citation is not evidence of a wrong product.

We did not establish intent, and nothing above should be read as an attempt to. We found a statistic credited to the wrong firm and a pasted URL. We found nothing that speaks to how either happened.

Our reading of the McKinsey article is a reading of the page as served. mckinsey.com returns no response at all to an automated request from our host, including for its own home page. We read the article through a rendering proxy on 27 August 2026, which returned the complete piece: four sections, all ten footnotes, the author line and the exhibit captions. The six exhibits themselves are images, and we did not read numbers inside the graphics. If a 60% figure exists inside one of those pictures, our search would have missed it — and the filing’s sentence would still be sourced to something McKinsey never wrote in prose.

We did not open the Bain research brief the press release points to. It may disclose everything the release omits. Our claim is about the release, because the release is what carries the number into circulation.

And the 60% may well be true. Nothing here tests it. What we would say about any single global zero-click figure, Bain’s included, is what the largest clickstream dataset we have covered says: within one study of 13.1 billion search events the rate runs from about 41% in Japan to 66% on US mobile. One blended number for that metric is the part to be suspicious of, whoever publishes it.

On the date. EDGAR’s “FILED AS OF DATE” for this accession is 25 August 2026, and that is the date we use. The raw acceptance timestamp in the same header is 24 August 2026 at 17:36 Eastern, after the close of that filing day, which is why EDGAR assigns it to the next one.

Sources

The filing and the Bain release were fetched and read in full on 27 August 2026 and both returned HTTP 200. Every quotation above is verbatim from the source named beside it.