Who clicks anymore?
Blue Ocean Strategy and the mechanisms of position loss it cannot see
ARTICLE
Bilal Farhat
10/18/20264 min read
Something has changed in the way people search, and it took me a while to name it. The question goes into the box the way it always did, and the answer comes back already finished... written out at the top of the page, with a handful of citations tucked inside it. The blue links are still there underneath. Fewer and fewer hands reach them. I kept wondering whether that small change in behavior was a small change in the world, or a large one.
Everyone arguing about AI and the web is armed with a feeling. Pew Research Center went and got a number. It tracked 68,879 Google searches made by 900 American adults in March 2025, real browsing, not a survey of what people think they do.¹
Where no AI summary appeared, people clicked through to a website on 15 percent of their visits.
Where a summary appeared, 8 percent.
And the links inside the summary itself, the citations that are supposed to carry you onward? Clicked on about 1 percent of visits.
I want to be careful with these numbers, because the people who own the page are careful with them too. Google disputes the study, the methodology and the query set both, and says the clicks it sends to the web have not fallen in aggregate and are higher in quality.² Other measurements exist, mostly from firms that sell search marketing services, and they disagree with each other about magnitude while all pointing the same direction. So the honest sentence is this one: every serious attempt to measure the gap finds a gap, and nobody agrees on its size.
Here is the part that took me the longest to think through. In the remedies phase of the United States search case, the court found that AI Overviews has "potentially strengthened" Google's position, with search queries up 1.5 to 2 percent since launch.³ That finding reads like a rebuttal of everything above. It is not. Both things are true at once. An interface can serve its users well, grow the platform that owns it, and still quietly narrow the road to the businesses standing behind it. Nobody has to build a worse product. Nobody has to do anything a user would object to. And still, somewhere behind the answer box, a company whose service works perfectly watches its customers stop arriving.
That is what happened, in the audited record, to one company in particular. On 2 May 2023, Chegg told investors that ChatGPT was affecting its new customer growth, and its shares closed down 48.41 percent in a day. By the end of 2025 its revenue had fallen from 617.6 million dollars to 376.9 million, 39 percent in a single year, and the segment exposed to search lost 43 percent while its other segment, same company, same balance sheet, lost 7.⁴ Chegg does not claim it was out-competed, and it does not claim a chatbot bundled its product away. It claims, in a lawsuit that is still undecided, that the road its customers traveled to reach it began answering their questions itself.⁵ That is the company's account, not a court's finding, and honesty also requires saying that Chegg's decline began in 2022, before AI Overviews existed. The clean part of the evidence is the acceleration, not the whole fall.
The old literature had a word for part of this. Disintermediation... the cutting out of middlemen. But the word was built for the opposite direction, for the airline selling tickets on its own website and cutting out the travel agent. What we are watching now is what happens after the middlemen won. Search became the layer through which we reach nearly everything, and that layer has started answering instead of passing us along. The strategy books have no chapter for it. Their advice is differentiate, and differentiation assumes a customer who can see the difference, which assumes a customer who faces a choice. The customer in the Pew data is not choosing badly. The customer is not choosing.
From where I sit, as a competition economist, this is the moment the problem stops being a strategy question and becomes a structural one, which is why three regulators and a court are now circling it. And one more honest thing. Every number in this piece is American. How this mechanism behaves where digital intermediation is built differently, across Europe, the Gulf, the wider MENA region, has not been measured by anyone, as far as I can tell. That open space bothers me more than the numbers do.
I tested all of this in a research paper that began as an attempt to prove something else and did not survive its own evidence. The data, the calculations, the research table and every question I asked are published with it, open to anyone. You can read it here.
The gap is measured, and its direction is not in dispute. What it does to the firms standing behind the page is the subject of the rest of this series.
Notes
Athena Chapekis and Anna Lieb, "Google users are less likely to click on links when an AI summary appears in the results," Pew Research Center, 22 July 2025.
Liz Reid, "AI in Search is driving more queries and higher quality clicks," Google: The Keyword, 6 August 2025.
United States v. Google LLC, No. 1:20-cv-03010 (D.D.C.), memorandum opinion on remedies, 2 September 2025.
Chegg, Inc., "Chegg Reports 2025 Fourth Quarter and Full Year Financial Results," 9 February 2026.
Chegg, Inc. v. Google LLC, No. 1:25-cv-00543 (D.D.C., filed 24 February 2025), docket.
About the author
Bilal Farhat is a competition economist and independent scholar. This paper is written in a personal capacity. The views expressed are the author’s own and do not represent those of any organization or firm with which the author is, or has been, affiliated.
Contact: bilal.h.farhat@gmail.com