Expect nothing from Google. A federal judge just confirmed.

Judge Amit Mehta dismissed Penske Media's antitrust case against Google on 30 September, and Chegg's with it, in a 41 page memorandum opinion. Penske had argued that search ran on an implied bargain: publishers let Google crawl their work for free, and Google sends readers back. Mehta granted Google's motions to dismiss in full. His reasoning fit in one sentence. "But an expectation is not an agreement. It is simply how a general search engine works."

Penske lost on the part nobody thought was in dispute. Not the traffic numbers. The deal.

The content for traffic bargain was never a bargain

Penske pleaded real damage. Its complaint put the click-through reduction from AI Overviews at 34.5%, said AI Overviews appeared on roughly 20% of searches surfacing its links between late 2024 and early 2025, and reported organic affiliate revenue down by more than a third from peak by the end of 2024. None of that was the problem. Mehta accepted the harm, conceding "the knock-on consequences to journalists, educators, and other online creators whose content Google takes and repurposes without compensation," then sent publishers to Congress and the regulators instead.

What killed the case was the missing counterparty. The opinion found that the publishers "failed to plead any actual agreement whereby Defendants promised to 'sell' Plaintiffs any specific amount of traffic," and that they "do not claim that this term arose from a meeting of the minds with Google, as opposed to their voluntarily acceding to cost-free crawling."

Read that last clause twice. So the arrangement was: you make everything, you give it away, and you receive whatever happens to be left over. And when there is nothing left over, the finding is that you agreed to this.

Every federal claim went with it, from reciprocal dealing to monopoly leveraging, and a final, appealable order came attached.

One percent of visits produce a click on a cited source

The underlying numbers are worse than the ones Penske pleaded. Pew Research Center tracked 68,879 Google searches from 900 US adults through March 2025. People who met an AI summary clicked a traditional result 8% of the time, against 15% when no summary appeared. And on 1% of visits, somebody clicked a source cited inside the AI answer itself. Google said the study "uses a flawed methodology and skewed queryset that is not representative of Search traffic" and that it has "not observed significant drops in aggregate web traffic."

Aggregate is doing a lot of work in that sentence. Chartbeat data covering more than 2,500 publisher sites, published by the Reuters Institute in January, put organic Google search down 33% globally and 38% in the United States between November 2024 and November 2025. The 280 media leaders the institute surveyed expect another 43% to go inside three years, and a fifth of them expect to lose more than 75%. Traffic arriving from ChatGPT is growing quickly. The report calls it a rounding error.

Jason Kint, who runs publisher trade body Digital Content Next, put the asymmetry plainly. Google, he said, has "an adjudicated illegal monopoly in search and is now swallowing up all of the journalism of humanity to power its AI products."

The most quoted case is HubSpot, where third-party panels estimated blog traffic down around 80%. HubSpot's SVP of marketing Kieran Flanagan said publicly that "the numbers aren't accurate, but the trend is," and attributed much of the fall to pruning more than 30,000 blog pages that were not converting. A half-conceded direction and a disputed size is roughly where every number in this category sits.

The floor was always zero

None of those figures is the story, because none of them is actionable. The ruling is, because it answers a question most revenue leaders have never put in writing: what entitles us to this channel? For organic search, the answer is now on the record. Nothing does.

We reported in August that Reddit's ChatGPT citations fell 86% in four days, with four parties disagreeing about whether it had happened at all. The ground under AI visibility moves without notice. What is new is that the recourse has now been tested and found not to exist.

That should land on anyone signing off a seven-figure GTM budget. The channels that produce the cleanest reporting are the channels where nobody has committed to anything. A dashboard records a volume that somebody else sets and can change without warning. There is no relationship anywhere in that number.

Set against that, the worst instrumented channel reads differently. We have argued that exhibitions take 40.8% of the average exhibitor's marketing budget and are still the hardest line to defend at budget time. That is a measurement problem. It is not an entitlement problem. When a VP shakes a hand at a dinner and agrees to a pilot, there is a named person on the other side who said yes, and no product update reassigns them.

So here is the falsifiable version. Over the next four quarters, teams that hold or grow their in-person budget will outperform teams that move the same money into AI visibility work, not because a room converts better, but because a room cannot be taken away without somebody telling you first. The first CFO to ask which channels we have an agreement in, rather than which channels we have a chart for, will reallocate faster than the market does.