Three named companies that disclosed real AI-search traffic losses

Most publishers never say what their AI-search losses actually are in a number anyone outside the company can check. Three exceptions exist, because a public earnings release or a traffic-panel analysis by a named research firm put a figure on the record. A fourth widely circulated claim, about CNN, does not survive contact with a primary source and is left out below rather than repeated.

Chegg: a 24% revenue drop the company put in writing#

Chegg’s own fourth-quarter 2024 results, published on its investor relations site on 24 February 2025, report total revenue of $143.5 million, down 24% year-over-year. This is not a third-party estimate: it is Chegg’s own disclosed financial result. CEO Nathan Schultz went further than the number itself, telling investors that Google’s AI Overviews had turned Google “from a search engine into an answer engine” that keeps traffic on Google’s own results page instead of sending it to sites like Chegg, and citing non-subscriber traffic that fell 49% in January 2025 compared with an 8% decline as recently as Q2 2024. Chegg has since sued Google over the same claim. This is the cleanest case in this piece: a real public company, its own disclosed number, and an on-the-record executive attribution to a specific Google product.

Business Insider: real traffic loss, a less clear-cut cause#

Press Gazette’s reporting, using Similarweb data, puts Business Insider’s monthly traffic at around 160 million visitors in 2020, falling to around 80 million in the first half of 2025 - a real decline of roughly half over five years. That is a verified, checkable number from a credible trade publication that tracks publisher traffic monthly. It is not, however, a clean before/after AI-search comparison: the five-year window spans Business Insider’s own editorial and paywall changes as well as the period when AI Overviews rolled out, and Press Gazette’s coverage of the same newsroom cuts does not isolate AI search as the sole cause the way Chegg’s own statement did.

HubSpot: a real collapse, but the algorithm did more than AI Overviews#

HubSpot’s organic traffic, tracked by Semrush and reported by Search Engine Land, fell from about 13.5 million monthly visits in November 2024 to about 8.6 million in December 2024 - a drop of roughly a third in a single month. That is a real, sourced number. What it is not is confirmation of the “AI killed 70-80% of HubSpot’s traffic” framing that circulates in SEO commentary: Search Engine Land’s analysis traces the December collapse mainly to Google’s December 2024 core update and spam update, compounded by HubSpot’s own pattern of publishing broad, SEO-first content - pages on topics like resignation letters and famous quotes - outside its actual product expertise. AI Overviews are named as a contributing factor in HubSpot’s own commentary on the episode, not as the credited primary cause in the reporting that has actual traffic numbers attached to it.

Why we care#

These are the exception, not the rule. A public earnings call or a named analytics firm’s monthly tracker is what makes a number here checkable at all. The overwhelming majority of sites affected by AI search changes have no equivalent public disclosure, so a claim about “publisher traffic” in general cannot borrow Chegg’s, Business Insider’s, or HubSpot’s specific number as proof of a universal effect.

Attribution is not automatic just because AI search changed in the same period. Chegg’s CEO made a direct, on-the-record causal claim tied to a specific product. HubSpot’s own reporting points mostly at an algorithm update and its own content choices, with AI Overviews as one factor among several. Treating every publisher traffic drop published since 2024 as an “AI Overviews did this” data point, the way the dropped CNN figure was being used before this check, is exactly the kind of unsourced leap this site does not ship.