ENTRY 025 · AI-OVERVIEWS · By Answer Engineered Research
+$14.6Mvs-$5.4M
Advertising Up 2%. Then Down 1%. The AI Overviews Sentence Survived.
People Inc. recast its 2025 fiscal year on 11 September. Advertising flips from a $14.6M increase to a $5.4M decrease. The AI Overviews line does not.
What People Inc. filed on 11 September
The filer is People Incorporated, formerly IAC Inc., formerly IAC/InterActiveCorp, CIK 0001800227. It owns PEOPLE, Investopedia, Better Homes & Gardens, Travel + Leisure and others.
The 8-K itself states its own purpose in one long sentence. Quoted verbatim, the company:
is filing this Current Report on Form 8-K (“Form 8-K”) to revise the sections entitled “Item 7. Management’s Discussion and Analysis of Financial Condition and Results of Operations,” “Item 7A. Quantitative and Qualitative Disclosures About Market Risk” and “Item 8. Financial Statements and Supplementary Data” contained in the Annual Report on Form 10-K for the fiscal year ended December 31, 2025 (File No. 001-39356) filed by the Company on February 20, 2026, and as amended on April 1, 2026 by the Amendment No. 1 on Form 10-K/A
And the reason, also verbatim:
digital portion of a legacy agency business that had previously been included within the People Inc. Print segment now reports to the D/Cipher management team within the People Inc. Digital segment.
Effective 1 January 2026, per the same filing. A business unit changed which segment it reports into, so the prior year gets re-presented on the new basis. This is ordinary, it is disclosed, and it is the whole stated reason.
The same sentence, two signs
Here is the original sentence, verbatim, on what moved the revenue line:
due primarily to increases of $46.6 million, or 21%, in Performance marketing revenue, net of intersegment eliminations, $32.9 million, or 28%, in Licensing and Other revenue and $14.6 million, or 2%, in Advertising revenue, net of intersegment eliminations.
And the recast, verbatim:
increases of $46.6 million, or 21%, in Performance marketing revenue, net of intersegment eliminations, and $32.9 million, or 28%, in Licensing and Other revenue, partially offset by a decrease of $5.4 million, or 1%, in Advertising revenue.
Set them side by side, one line per component.
| Revenue component, FY 2025 | Original 10-K, 20 Feb 2026 | Recast 8-K exhibit, 11 Sep 2026 |
|---|---|---|
| Performance marketing revenue, net of intersegment eliminations | Increase of $46.6 million, or 21% | Increase of $46.6 million, or 21% |
| Licensing and Other revenue | Increase of $32.9 million, or 28% | Increase of $32.9 million, or 28% |
| Advertising revenue | Increase of $14.6 million, or 2% | Decrease of $5.4 million, or 1% |
Two rows are identical to the dollar and to the percentage point. The third is an increase in one filing and a decrease in the other, for the same twelve months.
One more difference worth reading closely, because it is easy to miss. In the original, the qualifier “net of intersegment eliminations” is attached to the advertising figure as well as to performance marketing. In the recast, it appears once, on performance marketing only, and the advertising figure carries no qualifier. We are reporting that as a textual difference between the two sentences, not as an explanation of the sign change. The filing does not say the two are connected.
Where the dollars went
The segment tables are where the reclassification is visible, and they are worth doing the arithmetic on, because the arithmetic is the reason this is not a scandal story.
| FY 2025 revenue, in thousands | Original 10-K | Recast 8-K exhibit | Change |
|---|---|---|---|
| People Inc. Digital | 1,108,391 | 1,161,886 | +53,495 |
| People Inc. Print | 684,772 | 622,305 | -62,467 |
| Intersegment eliminations | (31,090) | (22,118) | +8,972 |
| Net effect | 0 |
Add the three changes: +53,495 minus 62,467 plus 8,972 is zero. Not approximately zero. Zero.
Total People Inc. revenue is identical in both documents: $1,762,073 thousand for 2025 and $1,777,229 thousand for 2024. No money appeared and no money vanished. Revenue moved between two segments, and the elimination line absorbed the difference.
What the filing does not do is connect that reshuffle to the advertising sentence. It explains the segment moves. It does not state that the segment moves are what turned a 2% advertising increase into a 1% advertising decrease. We are not going to assert the link for it. The dollars reconcile, the reclassification is disclosed, and the causal chain between them is not spelled out in either document.
The passage that barely changed
Now the part this field will keep citing. Here is the original passage in full, verbatim:
The increase in premium advertising was partially offset by lower programmatic revenue primarily due to lower impression volumes driven by a 5% decline in Core Sessions, due primarily to the impact of the increasing prominence of Google AI Overviews on Google search sessions, and an increased portion of impression volume consumed by premium advertising, partially offset by higher programmatic rates. The Company expects the increasing prominence of Google AI Overviews to continue to negatively impact Core Sessions and advertising revenue.
And the recast, verbatim:
The programmatic revenue decline was due primarily to lower impression volumes driven by a 5% decline in Core Sessions, due primarily to the impact of the increasing prominence of Google AI Overviews on Google search sessions, and an increased portion of impression volume consumed by premium advertising, partially offset by higher programmatic rates. The Company expects the increasing prominence of Google AI Overviews to continue to negatively impact Core Sessions and advertising revenue.
Be precise about what moved and what did not, because the sloppy version of this observation is already easy to write and it is wrong.
The lead-in was rewritten. “The increase in premium advertising was partially offset by lower programmatic revenue primarily due to” became “The programmatic revenue decline was due primarily to”. In the original, the programmatic decline is presented as a partial offset to an increase in premium advertising. In the recast, it stands on its own as a decline.
Everything after that point is identical. From “lower impression volumes driven by a 5% decline in Core Sessions” onward, including the entire second sentence, the two passages match word for word. That covers the whole causal attribution to Google AI Overviews and the whole forward-looking statement about it.
So the sentence most likely to be quoted out of this filing, the one that says the company expects AI Overviews to continue to negatively impact Core Sessions and advertising revenue, is the same sentence it was in February.
The sessions nobody re-measured
The operating metrics are identical in both filings.
| Operating metric, in millions | 2025 | 2024 | Change | % |
|---|---|---|---|---|
| Total Sessions | 9,546 | 10,664 | (1,118) | (10)% |
| Core Sessions | 8,602 | 9,062 | (460) | (5)% |
The recast adds a third year that the original did not carry: Core Sessions of 8,370 and Total Sessions of 10,813 for 2023.
That extra column is one more endpoint, not one more measurement. It does not test the attribution in either direction, and the decline the filing describes is still the 2025 against 2024 one. We are showing the 2023 figures because they are in the document and a reader is entitled to see them, not because a 2023 comparison settles anything the filing did not itself make.
One caveat that gets lost every single time this filing is cited: sessions are not clicks. Total Sessions and Core Sessions are session counts. Nothing in either document converts them into clicks, referrals or visits to any individual publisher property, and neither will we.
The OpenAI sentence, in both versions
One more line is word-for-word identical across both filings, and it belongs next to the AI Overviews passage rather than in a separate story. Verbatim:
The increase in Licensing and Other revenue was due primarily to improved performance of Apple News+ and content syndication partners and to the contribution of a full year of OpenAI revenue, a partnership which began in May 2024.
The same document that names Google AI Overviews as a cause of lost sessions credits a paid OpenAI licensing relationship for revenue growth. Both statements are in the original. Both statements survive into the recast unchanged. The $32.9 million, or 28%, increase in Licensing and Other is one of the two components that did not move at all between the two versions.
That is not a contradiction and we are not presenting it as one. It is the actual shape of the position a large publisher is in right now, stated by the publisher, in the kind of document where being wrong carries consequences.
What a recast actually checks
Here is the limit of this evidence, stated plainly, because the temptation to overreach is the entire reason posts like this go bad.
A sentence surviving a recast is not evidence that the sentence is wrong. It is not evidence that it is right either. It is evidence about what a recast checks.
A recast under a new segment structure re-derives dollar figures. Accountants re-allocate revenue between segments, re-run the eliminations, and re-present the tables and the narrative that describes them. That process has a specific scope. It touches the numbers whose segment assignment changed, and it touches the prose describing those numbers. It does not re-run a measurement of how Google search behaviour affected sessions, and there is no reason it would. Nothing about moving a legacy agency business between two segments would prompt anyone to re-test a causal claim about AI Overviews.
Which produces the thing worth knowing. A reader citing “People Inc. blames AI Overviews for a 5% Core Sessions decline” is citing a sentence that has now passed through a full re-presentation of the fiscal year without ever being re-tested, sitting in the same document as a revenue figure that did change sign. The survival of the one and the flip of the other are not the same kind of event, and only one of them tells you the filing was re-derived.
If you cite this filing, three things follow
Name the document and the date, not just the company. Two SEC documents from the same filer describe the same fiscal year’s advertising line with opposite signs. “People Inc. reported” is now ambiguous on its own. Cite the 10-K of 20 February 2026 or the 8-K exhibit of 11 September 2026, and say which one you read.
Treat the advertising line as version-dependent and the sessions line as not. The $14.6 million increase and the $5.4 million decrease are the same twelve months under two segment structures. Total Sessions and Core Sessions are identical in both. If your argument leans on the advertising figure, the version you picked is doing some of the work.
A sentence that survived the recast has not been verified by it. The AI Overviews attribution is unchanged since February, and that is an interesting fact about recasts rather than an interesting fact about AI Overviews. If you want the claim tested, it will have to be tested by something that actually measures sessions, which neither of these documents does.
Sources
- People Incorporated, Annual Report on Form 10-K for the fiscal year ended 31 December 2025, filed 20 February 2026. Every figure and quote labelled original above is from this document.
- People Incorporated, Form 8-K Exhibit 99.1, filed 11 September 2026. Every figure and quote labelled recast above is from this document.
- People Incorporated, Form 8-K, filed 11 September 2026. The reclassification language quoted above is from this document.
All three documents were fetched and read on 12 September 2026, and every quotation on this page was string-matched against the filed text rather than against a summary of it. No figure appears here that is not in one of those three documents.