What I Do When the Search Ends Without a Click

I would stop managing the company to organic clicks on queries an answer already completes. If the search ends on the answer, the buyer never reaches a page of ours unless the answer cites a page worth opening. I would manage to that citation, and I would keep the click rate as a consequence rather than as the plan.

The measurement is 93% of AI Mode Searches End Without a Click. That page is the deeper resource. This essay is how I would use it as a founder: what I would stop funding, what I would build instead, and which movement in the data would put the click back on the scorecard.

The figures that took the click off my scorecard

Semrush data, as reported for September 2025, puts the share of Google AI Mode searches that end without a click at 93% (summary of the Semrush figure). Bain had measured 60% zero-click searches earlier in 2025, in the same reporting. The later figure is the one that changed the plan. A channel that still sends some clicks can be optimized. A mode that ends 93% of searches before a click is a different object, and I would not hire against the remaining 7% as if it were the old results page.

The same roundup lines up the rest of the picture. Seer Interactive is cited for a 61% year-over-year decline in organic click-through on queries with AI Overviews, and for a 35% click-through lift for brands that are cited inside the answer relative to competitors that are not. Ahrefs, November 2025, is cited for AI Overviews on 99.9% of informational queries. Pew Research, July 2025, is cited for 1% of users clicking a link inside an AI Overview. I would keep the 35% line next to the 93% line. The click that remains is concentrated on being the source inside the answer. Ranking underneath an answer that 1% of people open is a weaker prize than it was when the results page was the whole product.

Google's economics did not follow publisher traffic down. Forrester reported Q1 2026 search revenue up 19% year over year (Forrester). A separate account of publisher traffic puts Google organic visits to more than 2,500 news sites down 33% globally and 38% in the United States between November 2024 and November 2025, with news executives anticipating a further 43% decline over the next three years (Smalk). The search business can grow while the click to a publisher shrinks. I would not wait for those two series to reconcile before I changed what we measure.

What I would stop doing

The reflex I would retire is the 2022 sequence: more posts, tighter titles, more links, a ranking report. That sequence assumed rank, then click, then conversion. AI Mode breaks it at the click. When 99.9% of informational queries carry an Overview, the person can finish without the article. Publishing another summary of someone else's data feeds the engine a page it can skip in favor of the original. I stopped writing that kind of roundup and started publishing claims that had a named source and a scope, because those are the pages an answer can cite without inventing the evidence.

I would also stop reading a ranking as proof of citation. Being third in the classic results and being the URL inside the answer are different events. I have written about how engines decide which sources to cite. The short version I would operate: original claims, a consistent entity across our site and third-party pages, and corroboration from a source we do not own. A guest-post link that nobody would defend in an edit meeting does not do the third job. A reporter who checked the claim does.

Time Magazine's reported response is the publisher version of the same move. It reduced Google's share of total traffic from 60% to 51% and grew ad revenue 22% year over year by building audience it did not have to rent from the results page (Smalk). I would not copy a magazine's ad model. I would copy the decision to stop treating the lost click as the thing to win back with more of the tactic that lost it. The account those writers give is that influence now allocates per citation. If a buyer asks which firm to hire and the answer cites a competitor, our analytics never see the buyer. The citation was the decision.

How I would reassign a quarter that was built on clicks

I would take the queries that produced organic pipeline last year and sort them into two piles before I touched the content calendar. Pile one is the query an answer now completes. Pile two is the query that still ends on a URL of ours. Semrush's 93% zero-click rate on AI Mode, September 2025 in the reporting I am using, is the reason pile one will be large. Bain's 60% earlier in 2025 is the reason I would not describe this as a one-month shock. Ahrefs' 99.9% of informational queries carrying an Overview, November 2025, tells me pile one is almost the whole informational set. Pew's 1% of users clicking a link inside an Overview tells me the leftover click is not a strategy. I would keep spending on pile two. I would stop writing pile one as if the click were the product.

Seer's two figures decide what "stop" means. Organic click-through on Overview queries down 61% year over year is the loss. A 35% click-through lift for the brand that is cited inside the answer, relative to a competitor that is not, is the remaining path. I would put the citation inside the answer on the goal line for pile one, and I would treat any click that follows as the 35%, not as a return of the old channel. E-commerce at a 3.2% AI override rate stays in pile two until that rate moves. I would write the exception down so a retail query does not get dragged into a services plan, and so a services query does not hide behind the retail exception.

The publisher numbers are the humility check, not a content idea. More than 2,500 news sites, Google organic down 33% globally and 38% in the United States from November 2024 to November 2025, executives expecting another 43% over three years. If that panel cannot hold the click, a startup blog running the same playbook will not hold it by trying harder. Forrester's Q1 2026 note, search revenue up 19% year over year, is the other half. The search business can grow while the visit to the publisher shrinks. Time's reported move, Google's share of traffic from 60% to 51% and ad revenue up 22%, is a publisher protecting a model I do not have. What I would copy is the refusal to spend the next quarter winning back a click the product no longer sells. Influence allocated per citation means a buyer who asks which firm to hire, and receives a competitor's URL, never appears in our analytics as a loss. I would rather see that loss in a saved transcript than miss it in a traffic chart that looks stable because the lost buyer never arrived.

The month I would run is still the short sequence, and I would staff it like a measurement, not like a campaign. Five buyer questions, saved with engine, date, and cited URLs. A map of what the cited competitor actually has on the page: a number, a name, a third party. One asset on our domain that states a claim with those properties and uses the same entity string everywhere else. One independent reference, a person who checked the claim, not a swapped post. Then the same five questions, same words. I would not add prompts because the first five were uncomfortable. Changing the question is how a team reports progress it did not earn. I stopped the roundup posts that summarize other people's data for the same reason. The engine can cite the original. Another summary gives it a reason not to cite us.

I would review the quarter on whether pile one's answers cite us, not on whether pile one's rankings moved. A rank with no citation is a vanity metric in a mode that ends 93% of searches before the click. The stack under that review is the source, the entity, and the corroboration, in that order. A guest-post link does not corroborate. A reporter who looked at the evidence does.

What I would do

I would take the five questions a buyer asks before a purchase and run them, once, on ChatGPT, Perplexity, and Gemini. I would save the answer, the cited URLs, and the date. If we are absent, I would not start with a technical audit of the blog. I would map, for each question a competitor wins, what that competitor has that the answer used: a number, a named person, a third-party page.

I would then build one asset against the highest-intent of those questions, on a domain we control, with the company and the founder named in a way a second source can repeat. In the following weeks I would try to earn one independent reference to that asset, from a reporter or an analyst who can look at the evidence. I would not count a swapped guest post. I would rerun the original five questions on the original wording after that reference exists. Changing the prompt and calling the new answer an improvement is how this work lies to itself.

The stack I would hang that sequence on is Machine Relations. The names of the events, a citation against a click, live in the glossary, so a report cannot rename a missing click as a "visibility win" without anyone noticing.

Scope: companies that sell information, expertise, or a service a buyer can shortlist from an answer. The reporting I am using puts e-commerce AI Overview rates much lower, at 3.2% in the same Semrush roundup, so a retailer whose demand is still a product-grid click is outside this decision. The trend described alongside that 3.2% is toward broader coverage. I would watch it. I would not pretend it has already arrived for every query class.

What would change my mind: a sustained fall in the 93% zero-click rate on the query class that produces our pipeline, with organic click-through back near the pre-Overview baseline, and with cited brands no longer showing the 35% lift. If the click returned as the main path, I would put rankings back on the scorecard beside citations. Pew's 1% in-overview click and the 61% organic CTR decline are the opposite of that return. While they hold, the plan is the source inside the answer, measured by rerunning the same questions, not by a ranking we can no longer cash.

FAQ

Does this apply to every query?

The Ahrefs figure cited for November 2025 is AI Overviews on 99.9% of informational queries, with commercial and transactional coverage described as expanding. The same reporting puts e-commerce override rates at 3.2%. I would apply the citation plan where our buyers ask informational and shortlist questions. I would not apply it blindly to a product query that still ends in a grid.

Can a cited brand still receive a click?

Yes. Seer Interactive's figure in this reporting is a 35% click-through lift for brands cited in the answer, set against a 61% year-over-year drop in organic CTR on Overview queries. I would chase the 35% by becoming the cited source. I would not chase the lost blue-link click with a larger content calendar.

What would the first month look like?

Five buyer questions, saved with their citations. A gap map where a competitor is cited and we are not. One page that states a claim with evidence and names the entity the same way everywhere else does. One independent reference to that page. Then the same five questions again. I would not call a four-week sequence a finished system. I would call it the start of a record I can compare.

Why would publisher traffic falling matter to a startup?

Because the publishers in the 2,500-site panel, down 33% globally and 38% in the United States from November 2024 to November 2025, had more SEO capacity than a startup blog, and they still lost the click. Google's Q1 2026 search revenue was up 19% year over year in Forrester's account. I would not bet the company on out-executing that panel at the old game. I would bet it on being the source the answer uses.