For most of paid search history, the funnel was a feature, not a worry. You could run a campaign on "what is a CRM" and a campaign on "buy Salesforce license," and both made sense. The first one cost less, captured someone early, and warmed them up. The second one cost more and closed. Paid search covered the whole journey, from the first vague question to the final purchase, and a good account manager treated that span as the point.
That span is being cut. The reason sits at the top of the results page: an AI Overview, a written answer that resolves the question before the searcher reaches a single link. When the page leads with an answer, the early, exploratory, informational query stops producing the click it used to. The transactional query, the one where someone wants to buy or book or get a quote, still does. The result is a channel that is not dying so much as contracting toward one end of itself.
This piece works through the evidence on both sides and asks the honest question a paid search decision-maker actually needs answered. Is search becoming a bottom-funnel-only channel, and if it is, what should change about how you run it?
Origin: how the answer climbed to the top of the page
Google did not arrive here suddenly. The company spent a decade lifting answers onto the results page through featured snippets and knowledge panels, but those handled trivia. The real shift started at Google I/O in May 2023, when Google announced the Search Generative Experience, an opt-in Labs experiment that put a generative answer above the blue links. SGE ran as a test for a year. Then in May 2024 it stopped being a test. Google renamed it AI Overviews and switched it on for all US users, with a stated goal of reaching more than a billion people by the end of that year. By October 2024 it was live in over 100 countries.
The structural change matters more than the timeline. An AI Overview is not another result. It is a block that sits above the results, often filling the first screen, and it answers the query in prose with a few citations along the side. Traditional organic listings move down. Some ad placements move down with them. And on an informational query, the searcher frequently gets what they came for without scrolling at all. Pew Research tracked the real browsing of 900 US adults and found that when an AI summary appeared, people clicked a traditional result on just 8 percent of those visits, against 15 percent when no summary was present, and they ended the session entirely 26 percent of the time.
For paid search this created a specific problem. The ad auction still runs. But the page it runs on now opens with content designed to make the next click unnecessary, and that content is concentrated on exactly the informational queries that used to feed the top of the funnel.
Present: the evidence cuts two ways
The honest version of this story has two findings that point in opposite directions, and you need both to read it correctly.
The first finding is the alarming one. When an AI Overview appears, paid click-through on that query drops hard. Seer Interactive, tracking client data, found paid click-through rate on AI Overview queries falling from 19.7 percent to 6.34 percent between June 2024 and September 2025, a 68 percent decline. The searcher reads the answer, satisfies the question, and never reaches the ad. On research-heavy queries this is real lost click volume, and it is not coming back.
The second finding complicates the panic. Seer's 2026 update, a larger study covering 53 brands and 296.9 million paid impressions from January 2025 to February 2026, found that paid click-through was actually the steadiest channel of the three across the whole period, far more stable than organic. More striking, being cited inside the AI Overview changed the math. On informational queries, brands cited within the Overview averaged a 15.74 percent paid click-through rate, against 11.19 percent when the brand was not cited. The earlier study put the same effect at a 91 percent higher paid CTR for cited brands. Seer's own analysts are careful here: they cannot prove citation causes the lift, since brands with stronger authority are both more likely to be cited and more likely to win clicks anyway. But the pattern holds every month. Appearing inside the answer, rather than only beneath it, is associated with a better paid result, not a worse one.
There is a third piece. Ads are no longer just above and below the AI answer. They are moving inside it. Google confirmed that text and Shopping ads from existing Search, Shopping, and Performance Max campaigns are eligible to appear within AI Overviews, in the US and a list of other English-language markets, and that advertisers cannot opt out. At Google Marketing Live in May 2026 the company introduced two AI Mode ad formats: Conversational Discovery Ads, which use Gemini to tailor creative to a specific query, and Highlighted Answers, which place an ad directly inside an AI-generated recommendation list. Ads inside AI Mode reportedly appeared in roughly a quarter of AI search results by early 2026, up from near nothing a year earlier, after testing peaked higher and then settled. Google's VP of global ads has said people click AI Overview ads at about the same rate as traditional search ads. Treat that as an interested party's claim, but the direction is clear: the answer box is becoming ad inventory, not just a wall that blocks ads.
The thesis, examined honestly
Put the pieces together and a specific argument forms. It is not "paid search is dying." It is "paid search is concentrating toward the bottom of the funnel."
The mechanism is query intent. AI Overviews do not appear evenly. They cluster on informational questions and largely skip the queries closest to a purchase. Seer's data shows AI Overviews triggering on roughly 36 percent of informational queries but only about 8 percent of commercial and 5 percent of transactional ones. Comparison and question-format queries are smothered, with Overviews appearing on 95.4 percent of comparison queries in one analysis. A query like "buy carbon-fiber pickleball paddle" or "same-day crown dentist near me" still mostly returns a classic page with classic ad slots, because an AI summary cannot complete the transaction and the searcher knows it.
So the part of paid search that loses click volume is the upper-funnel, informational part, the keywords a marketer ran to capture someone early. The part that holds is the high-intent, transactional, bottom-funnel part, where the searcher needs to leave the page to act and the ad is the fastest route. The DAC analysis of this shift put it neatly: AI may change how a journey starts, but not how it ends. Search, in this reading, is not shrinking into nothing. It is shrinking into a closer.
The case has a real weakness, and ignoring it would be dishonest. AI Overview coverage on commercial and transactional queries is rising fast. One Semrush-based analysis found AI Overviews on transactional queries jumping from about 2 percent to 14 percent and on commercial queries from 8 percent to nearly 19 percent over twelve months. If that trend continues, the protected zone at the bottom of the funnel gets smaller, and "bottom-funnel-only" might one day mean "almost nothing." The bottom-funnel concentration thesis is the best read of the current data. It is not a guarantee about 2027.
Future and impact: how to run paid search now
If the channel is concentrating, the response is to run it deliberately as a concentrated channel rather than pretend it still spans the whole funnel.
Value the transactional terms properly. The high-intent keywords are now doing a larger share of the channel's real work, and they should be funded and bid like it. Expect more competition there. As informational queries lose click volume, advertisers crowd onto the queries that still convert, and that pressure pushes cost per click up on exactly the terms you most want. The Search Influence analysis of 2026 paid search frames the strategic move plainly: shift weight toward high-intent keywords and conversion-ready language, the "demo," "pricing," "book," "near me" terms, and accept that upper-funnel keywords will simply return fewer clicks.
Reset expectations on informational keywords instead of cutting them blind. Some upper-funnel terms still earn cheap, useful clicks when no Overview fires. The mistake is reading their declining click-through rate as account failure. It is the page changing shape, not the campaign breaking. Decide which informational terms still pay and let the rest lose volume on purpose.
Optimize to be the cited brand. This is the single most useful tactic the data supports. If appearing inside the AI Overview is associated with materially higher paid click-through, then the content and authority work that gets a brand cited is now a paid performance lever, not only an SEO concern. The line between earning a citation and winning a paid click has blurred, and the teams that treat them as one effort will out-execute the teams that keep them in separate departments.
Watch presence, not only clicks. When an AI Overview pushes your ad down, your click-through rate can fall while your impressions hold. Falling clicks with steady impressions means you are still being seen; it is a different situation from disappearing, and it calls for a different response. Impression share and presence become first-class metrics, not footnotes. This is the same measurement problem we cover in the zero-click search era: the click is no longer a complete account of whether search is working for you.
There is also a tooling angle. Google is rebuilding the ad system for this world, and AI Max for Search is the clearest example: keywordless matching and generated copy aimed at the conversational, long-tail queries the keyword model could never catch. It is worth testing, because the queries it targets are exactly the ones AI search is reshaping. But a new campaign type does not resolve the funnel question. It just changes the controls you use to face it.
The honest close is that this is moving quickly and the endpoint is not settled. AI Overviews are spreading into commercial queries. AI Mode is taking ad formats Google is still inventing. The bottom-funnel concentration thesis is well supported by today's evidence and could be overtaken by next year's. What a paid search leader can act on now is the direction, not the destination: informational search is being absorbed into the answer, transactional search still converts, and the channel is worth running as the closer it is becoming rather than the full-funnel tool it used to be.
Council summary
This post argues that AI Overviews are not killing paid search but concentrating it toward the bottom of the funnel: informational queries lose click volume to the answer box, while transactional queries, where a searcher still needs to leave the page to buy or book, keep converting. The council verified every statistic against primary sources, including the Pew browsing study (8 percent click rate with an AI summary against 15 percent without), Seer Interactive's September 2025 and 2026 updates (the 68 percent paid CTR drop, the 296.9 million paid impressions, and the cited-brand figures of 15.74 against 11.19 percent), the Semrush finding that AI Overviews on transactional queries rose from about 2 percent to 14 percent, and the Google Marketing Live 2026 ad formats. The thesis is presented with its honest weakness intact: rising Overview coverage on commercial queries could shrink the protected zone. The practical takeaway for a paid search leader is to fund transactional terms harder, reset expectations on informational keywords rather than cutting them blind, treat AI Overview citation as a paid performance lever, and measure presence, not only clicks.
Comments