TechCrunch Called Google Search Over. Three Query Types Didn't Get the Memo.

TechCrunch's May 19 piece declared the ten blue links dead and Google Search "over." The headline obscures three query types where Google's AI rollout has the smallest impact on click-throughs: branded searches, high-involvement transactional searches, and YouTube SERPs. A Search Engine Land UX study found 69% of AI Mode sessions on service searches still resulted in a website visit.

The obituary is real. The application is sloppy.

Sarah Perez published the piece the day Google's I/O announcements landed. The thesis: AI Overviews now serve 2.5 billion monthly users, AI Mode is at 1 billion, and Liz Reid's new "intelligent search box," generative UI, and 24/7 information agents will "further decimate Google referrals to publishers."

That part is correct. The implication, that all of Google Search is going the way of news publisher referrals, is where it falls apart. A 58% CTR drop on the top-ranking page when AI Overviews appear is brutal. But that average is hiding a wide variance across query types, and the averages are doing a lot of work that the publishers reading the obituary are not.

Reid didn't say every search is becoming AI Mode. She said the search box is. There is a difference, and the difference is the entire content strategy decision for the next six months.

What still works under AI Mode

Three query types are not following the publisher curve down.

Branded searches. AI Overviews appear on under 5% of branded SERPs. On the queries where they do appear, branded CTR seems to rise rather than fall. That makes intuitive sense. When a user searches "Notion pricing" or "Allbirds returns," Google has zero incentive to summarize away the destination the user has explicitly named. Brand authority is the moat AI has not crossed, because the user already supplied the answer. From what I've seen, this is the single most underrated reason to keep investing in brand search ads and brand authority work in 2026. Most teams treat branded query spend as a defensive line item; in this SERP environment it is closer to an offensive one.

High-intent transactional searches. A Search Engine Land study by Liam Wright tracked 52 U.S. and Canadian participants through nearly 22 hours of AI Mode sessions, focused on high-involvement service searches like dentists, dermatologists, and cosmetic providers. The result: 69% of AI Mode sessions resulted in a click to a website. Only 27% of participants felt ready to decide based on the AI summary alone. On average, they checked 3.7 results per session. Not even close to a zero-click world. AI front-loads the comparison; humans still drive the buy.

Semrush data hints at why Google is going gentle here. AI Overviews used to appear on roughly 2% of transactional queries; that has climbed to around 14%. Still low. Google appears to be guarding the auctions where the click is monetizable, because that is where the ad revenue lives.

YouTube SERPs. Google's AI Overviews increasingly pull video carousels from YouTube for product and location queries. "Best noise-cancelling headphones." "Museums to visit in San Francisco." "How to install a doorbell camera." If you have a YouTube channel with even modest depth on transactional or how-to terms, the carousel is a fresh distribution surface inside the AI answer, not a sidebar to it. Adweek's coverage of the original carousel test flagged this for "product and location searches" specifically, which are exactly the queries that still convert.

The non-obvious read

Most SEO publishers are writing this as a defensive piece. How to survive zero-click. What the data actually says is that Google is selectively cannibalizing one specific revenue model: informational content that exists to capture top-of-funnel ad impressions. The publisher business of the late 2010s.

What it isn't doing, at least not yet, is cannibalizing the business of a brand the user already trusts, a service the user is comparing for a decision worth more than a few hundred dollars, or a YouTube channel that ranks for the query.

To be fair, AI Mode's overall zero-click rate sitting around 93% is real and ugly. I'm not arguing the obituary is wrong. I'm arguing the obituary is being applied indiscriminately, and businesses with the wrong content mix are reading themselves into a panic when the actual question is which slice of their traffic they're losing, and which slice was never really theirs to begin with.

How to figure out which slice you are losing

Open Search Console. Pull the last 90 days. Bucket the queries into three groups.

1. Branded (the query contains your company or product name). If these clicks are stable or rising, the brand moat is holding. Funnel more spend there.
2. Transactional (contains "buy," "price," "near me," "review," "vs," or a specific service name). If stable, you're fighting the right fight in a healthy auction. If they're dropping faster than the site average, you've got a paid-vs-organic mix problem that AI Mode is amplifying, not a search-is-dead problem.
3. Informational (everything else). Assume these are the losses. From what I've seen, a 30-50% CTR drop on this bucket is roughly the new normal when AI Overviews fire, and there is no SEO playbook that gets the clicks back. The play is to convert these into YouTube videos, treat them as zero-click brand exposure, or kill them.

The exact split varies wildly by site. Condé Nast publicly admitted Google traffic dropped to 25% of its referrals and stopped pretending the old playbook works. A B2B SaaS site I've poked around in is barely down 8%, because almost everything they rank for is branded or comparison-driven. Same algorithm; very different outcomes.

The bet I would make for the next six months

Branded query investment, including the brand search auction in Google Ads, is the cheapest ROAS most teams aren't fully funding right now. Transactional SEO and PPC keep working, with sharper auction competition as informational budget reallocates downstream. YouTube optimization stops being a "video team thing" and starts being part of SEO, because the carousel is the new featured snippet for half the queries that used to send blue-link traffic.

What I'd stop doing: writing more 1,500-word listicles for informational keywords. The math hasn't broken for evergreen practitioner guides yet, but the thin top-funnel content factory is broken, and pretending otherwise is just delaying the inevitable rewrite.

The obituary writers are right that something ended. They're wrong that everything ended. If you only read the headline, you'll cut the wrong line items, and keep funding the line items that are actually leaking.