Publishers Diversify Away From Google Search Traffic After AI Overviews Bite

Full analysis

Google's AI Overviews, the AI-generated answer boxes that sit above the blue links and let people get their answer without clicking through, are now triggering on roughly half of the queries relevant to Ziff Davis, up from about a third the quarter before. Publishers are responding by scattering their bets: The New York Times into video, News Corp's Dow Jones into subscriptions at 81% of revenue, Gannett's USA Today Co. into newsletters and social video, Ziff Davis into apps and licensing, BuzzFeed into a cheaper contributor model. USA Today Co. president Kristin Roberts put it flatly: "We're not building our future on search, even in our strongest categories."

What's actually being decided here isn't a publisher story. It's a supply-quality story for everyone who buys and sells open-web impressions. If high-intent, search-referred readers are being intercepted before they land on a page, the audience that programmatic buyers thought they were buying is thinning. This is a Type 1 shift for the open web: hard to reverse, because Google isn't turning Overviews off. The forcing function is already here, showing up in earnings guidance right now.

The Market Analyst. Follow the impression math. Open-web programmatic players like Magnite and PubMatic (the pipes that sell publisher ad space in real time) live on volume from search-referred pageviews. Fewer clicks means fewer impressions means lower gross spend running through those pipes. That pressures their growth story. On the other side, the winners are the identity and clean-room vendors: LiveRamp, ID5, the players who help publishers turn logged-in, newsletter, and app audiences into sellable first-party segments. For a generalist: Google is keeping the reader on Google, and the value that used to flow to publishers and their ad partners flows back to Google's own ad stack. That last part is the cannibalization Alphabet hasn't reconciled out loud.

The Skeptic. We have seen this movie. In 2018, Facebook Instant Articles cratered referral traffic and every publisher swore off platform dependence. Google traffic came back and everyone quietly re-optimized for it. Query interception is not revenue loss until you prove those clicks were monetizing above the floor price, and a lot of informational search traffic never did. BuzzFeed's contributor model is a cost cut. Calling it a strategy is generous. The blunt read: some of this is investor-relations language to dress up soft guidance. If subscription and newsletter revenue can't replace programmatic CPMs at volume, and below NYT and WSJ scale the math is brutal, Google dependency reasserts inside 18 months.

The Operator. Reprice the traffic mix now, not next planning cycle. If half the relevant queries trigger an Overview, the CPMs tied to your search-referred audience are degrading quietly, because that cohort was your high-intent inventory and it's being skimmed before arrival. The break shows up in mid-funnel display around 90 days out: advertisers bought against "engaged, search-referred readers" and that cohort is thinning while your rate card pretends it isn't. Newsletter and direct traffic don't carry the same contextual signal at scale yet, so you can't just swap one for the other on the sell side. Revenue ops and audience development need to be in the same room this month.

The CFO. The diversification everyone's cheering is a margin story, and it's ugly for most. Programmatic CPMs are high-margin: the pageview exists, the ad renders, cash drops. Building newsletters, apps, subscriptions, and video is capital and headcount up front with a long payback. NYT and Dow Jones can afford the runway because subscriptions already carry them. For a mid-tier publisher, you're spending real money to replace a revenue line that was nearly free at the margin, and the replacement monetizes at a lower rate per user until scale arrives, if it arrives. BuzzFeed's contributor shift tells you which side of that math most players are actually on.

Where the council splits

Two disagreements matter. First, the Skeptic versus the Market Analyst on whether this is structural. The Skeptic says query interception isn't revenue loss and Google always comes back. The Market Analyst says the impression volume feeding the open-web pipes is genuinely shrinking, and that's a durable drag regardless of publisher messaging. They can't both be right about Magnite and PubMatic.

Second, the CFO versus the Strategist read implied in the diversification cheer. Owned audiences are the right long-term moat. But the CFO points out that only a handful of publishers can fund the transition, and the rest are trading a free high-margin line for an expensive low-margin one. "Diversify" is a clean word for a brutal capital decision.

What it hinges on

Three things decide this. One: were the search-referred clicks now being intercepted actually monetizing above the floor, or was a chunk of it low-value informational traffic Google is welcome to keep? Two: can first-party audiences carry the contextual buying signal that made search-referred readers premium, or do they sell at a discount for years? Three: does open-web impression volume fall enough to show up in the growth rates at Magnite and PubMatic.

The council leans toward this being real but uneven. The open web loses volume, identity vendors gain relevance, and the split between publishers who can fund owned audiences and those who can't widens. What to verify before acting: pull your own search-referred CPMs against direct and newsletter CPMs, segment by intent, and see whether the intercepted traffic was ever paying. That number tells you whether you have a revenue problem or a vanity-traffic problem.

Prediction: In their next earnings prints (Q3 2025/2026, roughly Oct-Nov 2026), at least one of Magnite or PubMatic will explicitly cite AI Overviews or declining publisher search traffic as a headwind to open-web display volume.

Confidence: Medium. The volume drag is mechanical, and management gets asked directly on calls.

Why: Ziff Davis is already reporting AI Overviews on half its relevant queries, and search-referred pageviews are the raw material open-web programmatic pipes resell, so fewer clicks mechanically means fewer impressions flowing through Magnite and PubMatic. Analysts on those calls will ask about AI's effect on supply, because it's the obvious question this cycle, and management typically names a headwind once it's undeniable rather than absorb the "why did you miss" narrative silently. The opposite outcome, total silence, is less likely because the interception is now big enough and public enough that dodging it on a call reads as evasive.

Revisit by 2026-11-30: We're right if either company names AI Overviews or falling publisher search traffic as a supply or display-volume headwind on its next earnings call. We're wrong if both attribute open-web softness only to macro, pricing, or CTV mix shift with no mention of AI-driven search interception.