Large language models are pushing the web toward zero clicks, according to an analysis published by UCLA Anderson Review — a shift that would extend the long-running decline of organic search referral traffic beyond anything traditional SERP features have caused so far.

The UCLA Anderson Review piece frames generative AI chatbots and AI-assistant answers as a structurally different threat to publisher economics than featured snippets, knowledge panels, or other classic search page features. Where those elements suppress some clicks while still leaving the user on a search results page, an LLM delivers a synthesized answer inside its own interface. The user never reaches a publisher's site at all.

That distinction matters for anyone who runs a content site, an affiliate business, or a publishing operation funded by advertising. The zero-click problem is not new — search engines have absorbed an increasing share of queries on their own results pages for years. But the Anderson analysis positions LLMs as an escalation: the click that featured snippets left on the table is the click a chatbot answer eliminates outright.

Why is this different from featured snippets?

Search results pages that answer a question directly still expose the user to ten blue links, ads, and brand visibility. A conversational AI answer collapses that entire page into a single response. If the model's answer satisfies the query — and for definitional, how-to, and comparison questions it increasingly does — the publisher's URL appears, at best, as a citation the user may ignore.

For site owners, the practical consequence is that rankings themselves may stop translating into traffic. A page can be the model's underlying source and still receive no visit, no ad impression, and no conversion path.

Which site types face the most exposure?

The analysis's logic points to clear vertical exposure patterns:

• Informational publishers whose revenue depends on ad impressions per visit lose the entire monetization event when the answer is consumed in-chat.
• Affiliate and review sites face the same squeeze if product recommendations are synthesized without a click-through to the merchant link.
• Reference, how-to, and definitional content — the categories most easily summarized — are the first to lose traffic.
• Sites whose content requires transactional completion (booking, checkout, tool use) retain a stronger click incentive, because the LLM cannot finish the task for the user.

What does this mean for the traffic model?

The Anderson analysis joins a growing body of commentary arguing that the web's default unit of consumption is shifting from the page to the answer. If that holds, the publishing model built on search referral volume — publish, rank, monetize the visit — erodes at its foundation. Publishers would need either direct relationships (newsletters, apps, subscriptions) or new economic arrangements with AI platforms to replace lost referral revenue.

It is worth separating what is established from what is projection. The direction of travel — more answers consumed without clicks — is supported by the analysis and by observable behavior in chatbot products. The precise pace and the ultimate share of queries that go fully clickless remain open questions, and the Anderson Review piece itself is an analytical argument rather than a ranking-event observation.

What should site operators watch next?

Monitor the gap between rankings and sessions in analytics: pages that hold positions but lose clicks are the earliest signal of answer-layer displacement. Watch how AI platforms handle attribution and citations as they mature, and whether licensing or referral arrangements emerge between model operators and publishers. The zero-click trend the UCLA Anderson Review describes will show up first in informational queries — and the sites that measure that gap now will see the shift coming before it shows in revenue.