Paid search was never going to save you from AI Overviews

The narrative for the last eighteen months has been clean and comforting. Organic clicks are being eaten by AI Overviews. Paid clicks are growing. Google's total revenue keeps breaking records. Therefore the demand didn't disappear — it just moved down the page, from the blue links to the sponsored slots, and if you're smart you'll follow it there.

Tinuiti's Q2 2026 benchmark report just made that story significantly harder to tell.

Paid search spending growth on Google slowed to roughly 14% in Q2, down from the 15% Google saw in Q2 2025 and softer than the acceleration advertisers had been enjoying through Q4 and Q1. YouTube slowed. Instagram slowed. Only Facebook rebounded, and Amazon's numbers are a Prime Day artefact. The story of paid absorbing organic's collapse held for about four quarters. It's now visibly wobbling.

This matters more than a single quarter's benchmark suggests, because a lot of 2026 budget decks — the ones being presented to UK boards right now — are built on the assumption that the paid line item will keep compounding at the rate it did in 2025. It won't. And the reason it won't tells you something structural about where search is going that most agencies aren't ready to explain to their clients.

The absorption thesis was always half a theory

Here's the version of events most of the industry has been running with. AI Overviews expand. Organic CTR craters — Pew put the click rate at 8% with an AI summary versus 15% without, and Ahrefs estimated a 58% CTR drop for the top-ranking page when an Overview appears. Users still have commercial intent. That intent has to land somewhere. Paid search — sitting at the top of the SERP, above the AI summary in most layouts — becomes the new default landing point. Advertisers spend more, click volume grows, everyone gets to keep their job.

The demand didn't relocate — some meaningful portion of it evaporated.

The problem with this thesis was always the middle step. Pew's own data flagged it: users seeing an AI summary were *more* likely to end the session without clicking anything at all. Not shift the click from organic to paid. End the session. The demand didn't relocate — some meaningful portion of it evaporated.

The reason paid search still grew through late 2025 and early 2026 wasn't that Overviews were funnelling clicks to sponsored listings. It was that commercial-intent queries — the ones where paid actually competes — were still throwing off click volume, and advertisers were bidding harder on a shrinking but qualified pool. Higher click volume, marginal CPC growth, more total spend. That worked while the base of commercial queries held up.

Q2 is the first quarter where you can see the base starting to soften.

What the deceleration actually shows

Read Tinuiti's numbers carefully. Google paid search spending grew 14% in Q2 2026 against a Q2 2025 comparison that had itself grown 15% on top of a 9% Q1 2025. The baseline is now genuinely tough. This isn't collapse. It's the end of easy comparisons.

Paid search was compounding against a soft base. That base is no longer soft.

The CPC number is the interesting one. Average search CPC increased just 1% year-over-year. Most of the spending growth came from higher click volume, not higher prices. That's the opposite of what you'd expect if the AI-Overview-to-paid-search absorption thesis were the primary driver. If organic demand were reliably being redirected to sponsored slots, you'd expect competition for those slots to intensify — CPCs would rise sharply, spending growth would be driven by price, and click growth would flatten.

What Tinuiti is showing is the opposite pattern. Click growth is doing the heavy lifting. Prices are barely moving. Which suggests advertisers are still finding new inventory to bid on — new query types, new placements, new formats like the anchor-text hyperlinks Google is currently testing in ad descriptions — rather than paying up to defend against fierce new competition on existing terms.

That's a fine story for growth in an expanding market. It's a nervous story if you thought paid search was going to be the mechanical beneficiary of AI Overviews.

The compression is happening on both sides

Zoom out from the Tinuiti data for a moment. iPullRank's zero-click analysis of 13 billion searches, published just last week, showed something that reframes this whole conversation: zero-click isn't a single number, it's a distribution. Some searches — informational, factual, "how tall is" — are effectively closed loops now. Google answers, session ends, nobody clicks anywhere. Other searches — transactional, comparative, high-consideration — still generate clicks reliably.

The commercial queries that fund paid search sit disproportionately in the second bucket. That's why paid search kept growing even as aggregate zero-click rose. Advertisers weren't competing for the queries that were being absorbed. They were competing for the queries that never were.

But the absorption is now moving upmarket. AI Mode and AI Overviews are getting better at handling comparative and transactional intent — the "best CRM for small business" query, the "cheapest flights to Lisbon in October" query. These are queries that used to reliably generate paid clicks. Some portion of them are now ending in an AI summary with citations, and the click — if there is one — goes to the cited source, not the paid slot.

That's the compression Q2 is starting to reveal. Not that paid is being replaced by organic clicks, but that the pool of queries where paid can compete for a click at all is starting to shrink at the top end.

What this means for how you plan H2

Most of the H2 planning I've seen from clients this quarter assumes paid search will keep doing the heavy lifting while organic gets figured out over 2027. The Tinuiti data makes that assumption expensive.

Three things follow from it.

The first is that CPC modelling built on 2025 assumptions is likely wrong in both directions. Some commercial verticals will see CPCs rise sharply as the pool of biddable queries contracts and competition intensifies on what's left. Others will see CPCs soften as query volume declines faster than advertiser demand adjusts. If your paid team is running the same efficiency model they were in Q1, they're going to be surprised by their own numbers in Q4.

The second is that the pressure to defend brand terms is going to increase. When AI Overviews cite a competitor for a query where your brand used to rank, paid brand defence stops being optional. This is going to eat budget that used to go to prospecting, and it's going to happen quietly — most teams won't notice until the annual reconciliation.

The third, and the one most agencies won't tell their clients, is that the case for reinvesting in organic brand-building just got stronger, not weaker. If paid search growth is decelerating because the base of biddable commercial queries is shrinking, then owning the *un*-biddable ground — the brand awareness, the entity recognition, the citations in AI summaries themselves — is where the marginal return lives. Brand is the only compounding asset in AI search, and the paid line item was never going to substitute for it.

The counterargument, taken seriously

The strongest version of the "paid will keep growing" case runs like this. Google has enormous incentive to keep the ad business growing. They're rolling out new placements constantly — anchor text hyperlinks in ad descriptions, new formats in AI Mode, Shopping expansion, YouTube integration. Every new placement is new inventory. As long as Google can manufacture new places to sell ads, the paid growth line stays intact regardless of what happens to organic.

I think this is true in aggregate for Google's business and probably wrong for individual advertisers. New ad formats grow Google's revenue. They don't necessarily grow *your* ROAS. If AI Mode ads become the dominant new format, the advertisers who win are the ones who figure out how to compete inside a conversational surface — and the mechanics of that are utterly different from bidding on keywords. Media efficiency in a manufactured-inventory environment tends to fall, not rise, for the median advertiser. Google's revenue chart keeps going up. Your CAC also keeps going up.

The steelman is that paid keeps growing. The rejoinder is that "paid keeps growing" and "your paid programme keeps working" are two different sentences.

What honest H2 planning looks like

If I were building an H2 forecast right now, I'd do three things differently from the standard 2025 template.

I'd stop modelling paid search growth as a straight-line continuation of Q1's rate, and I'd start modelling it as a distribution — some verticals accelerating, others decelerating, and I'd force the paid team to declare which side of the line they think they're on and why. If they can't, that's the finding.

I'd treat the AI Overview reports in Search Console as inputs to media planning, not just SEO reporting. Which of your commercial queries are being absorbed into Overviews? That's the leading indicator for which paid queries are about to see either a CPC spike or a volume decline.

And I'd redirect a portion of the "efficiency budget" — the money that used to go into wringing another 5% ROAS out of a mature Google Ads account — into brand and citation-earning activity. Not because it's morally superior. Because the mechanical return on the efficiency budget is deteriorating and the return on brand is compounding.

The paid absorption story worked while it worked. It stopped working in Q2. Anyone who tells you otherwise between now and December is showing you a chart with the y-axis cropped.

The demand didn't move from organic to paid. Some of it did. Most of it just left. And the businesses that spend the second half of 2026 pretending otherwise are going to spend the first half of 2027 explaining variance to their boards.