Google Built a Toll Booth Inside AI Mode. Stop Paying the Toll Expecting a Parade.

A new study says almost a third of searches now show ads inside Google's AI Mode, and the one thing everybody assumed those ads would buy you (a citation, an organic boost, a little AI halo) mostly isn't happening.

A client called me last month asking if she should start bidding harder in Google's AI Mode because her competitor's brand kept "showing up in the AI answers." She wanted to know if paying for placement there would get her cited more, ranked higher, trusted more by the algorithm itself. I told her I didn't know. Turns out I wasn't the only one guessing.

You see, that's the exact question a new study just answered, and the answer is no. Paying for the ad doesn't buy you the thing everyone assumed it would.

What happened

A study covered by Search Engine Land found that Google's AI Mode now shows ads on nearly 30% of all queries, and the higher the CPC on a search term, the more likely an AI text ad shows up next to (or inside) the AI-generated answer (Search Engine Land). The part that should stop you mid-scroll: paid placement in AI Mode rarely translated into more citations or better organic rankings. Buying the ad got you the ad. It didn't get you the AI's trust, or a boost in the "real" answer sitting above or below it.

That lands the same week Digiday reported that publisher ad supply on the open web fell by up to 40% in Q2 of 2026, according to U.S. and U.K. benchmarking data from Ozone shared exclusively with Digiday (Digiday). Zero-click, AI-driven search is drying up the traffic that used to flow to news sites and blogs, which means the inventory brands used to buy in bulk is shrinking fast.

Two stories, one week, same root cause. AI search is rerouting where attention goes, and Google is charging admission at the new door while the old doors quietly close.

Why this matters

For years, the pitch on SEO and content was simple: rank well, get cited, earn organic trust, and paid media was the accelerant on top of that. This study suggests AI Mode doesn't work that way. It's not accelerating your organic presence. It's a separate paid lane that runs parallel to the organic answer, and money spent in one doesn't refill the other.

Meanwhile, the Digiday numbers say the open web itself, the place where display and programmatic dollars have lived for two decades, is shrinking as a delivery mechanism because fewer people click through past the AI's own answer. If that number holds up (and a 40% single-quarter drop from one vendor's benchmarking data deserves a second and third data point before anyone treats it as settled), it means the inventory pool advertisers have relied on is not just fragmenting, it is contracting.

Put those together and you get a genuinely uncomfortable picture for anyone still running budgets like it's 2022: the traffic is drying up in the old channel, and the new channel takes your money but won't lend you any credibility.

My take on all of these

Here's the part that gets me. Everybody in this business has spent thirty years assuming that if you pay Google enough, something good eventually leaks over into the free stuff. Paid search helps quality score. Ads near organic listings build brand recall that lifts click-through on the organic result below it. There was always a little bit of osmosis.

This study says AI Mode doesn't leak. It's sealed. You pay, you get the ad slot, and the AI answer above it goes on citing whatever it was already going to cite, based on whatever signals it actually cares about (which, near as anyone can tell, isn't your media budget).

I don't think that's Google being sneaky exactly. I think it's Google being honest for once about what advertising actually is: a rented spot, not an investment in reputation. The industry just never wanted to hear it stated that plainly because the whole SEO-plus-paid-search bundle made more sense as a story where everything reinforces everything else.

Combine that with publishers losing up to 40% of their ad supply and you've got a media landscape where the safe, familiar places to spend money are either shrinking or refusing to give you the halo you thought you were buying. That should scare you a little. It scares me a little. I don't have a tidy answer for where the next decade of ad dollars land, and anyone who tells you they do is selling something.

What to do now

Stop budgeting for AI Mode placement as if it's an SEO play. It's a media buy. Judge it on direct response, not on hoped-for organic lift, and measure it separately from your content and SEO line items so you're not fooling yourself about what's driving what.

Audit your programmatic and open-web spend this quarter, not next year. If Ozone's numbers are even directionally right, the premium CPMs you're paying for shrinking inventory need a hard look against where actual eyeballs are going now.

And ask your agency the blunt question: which budget line is buying attention, and which one is buying hope? If they can't answer that fast, that's your answer.

Also in AI and advertising this week

• Attribution's credibility problem. AdExchanger asked whether falling match rates make effective attribution "just an illusion," featuring Cimin Ahmadi Cohen of Idea Peddler questioning whether the dashboards clients obsess over were ever measuring what people thought they were (AdExchanger).
• First-party data's unfinished job. Another AdExchanger column argues brands that spent years building CDPs and loyalty data still hand media buying decisions to platform algorithms trained on someone else's data, treating first-party data as a suppression list instead of a real buying strategy (AdExchanger).
• Reddit's data deal dilemma. Digiday reports Reddit is internally debating whether licensing its community conversations to AI companies could undercut the very thing its ad business sells: unique, human conversation (Digiday).

Sources

• Search Engine Land: Google AI Mode ads reach nearly 30% of queries
• Digiday: Publisher ad supply fell by up to 40% in Q2
• AdExchanger: With Match Rates Falling, Is Effective Attribution Just An Illusion?
• AdExchanger: First-Party Data Is About More Than Retargeting And Audience Suppression
• Digiday: Reddit questions if AI data deals could hurt its ad business