Ads hit 33% of AI Mode queries, but advertisers rarely get cited

Google charges brands to appear in AI Mode. It just does not promise to cite them.

Search Engine Journal reports that a new analysis of Google AI Mode found text ads appearing on roughly one in three commercial keyword searches. That figure alone is not the story. The story is what sits alongside those ads: the sources Google's AI actually cites when generating its answers are, with notable regularity, not the brands paying to be there. Two separate things are happening in the same interface, and most marketing teams are treating them as one.

The paid-organic split is now a paid-invisible split

Classic search created a clean, if imperfect, bargain. Buy the top slot and you dominate the page. Organic optimisation could challenge you, but the layout made your presence visible and your click measurable. AI Mode dissolves that bargain. A brand can run a text ad against a commercial query, win the auction, and still watch its competitor's content, or an independent publisher's, get woven into the generative answer a metre above its ad unit. The user's attention goes to the answer. The ad sits beneath it, carrying the brand's budget and very little of its authority.

This is not a hypothetical edge case. If ads appear on roughly a third of commercial queries in AI Mode and advertisers are rarely the cited sources, the structural mismatch is systematic. It scales with spend: the more aggressively a brand buys keywords in AI Mode, the larger the gap between its paid presence and its cited presence becomes.

For sectors where authority and trust are constitutive of the product itself, the gap carries real cost. A financial services firm that runs ads against queries about retirement planning, but whose research is absent from the generative answer, is paying to frame a conversation it is not having. A multilateral institution promoting a policy report gets no citation credit from a paid line that points to its landing page, while a think-tank that never bought the keyword gets quoted in the synthesised response above. Visibility in the ad slot and credibility in the answer are being priced separately, whether or not brands have noticed.

What determines the cited source is not the winning bidder

Google's generative answers draw on signals closer to traditional organic ranking and, increasingly, on what the model treats as authoritative for a given query type. Primary research, structured data, named expert sources, editorial consistency and domain trust all appear to influence which sources surface in the synthesised text. Paid placement influences none of those factors directly.

The implication for B2B brands with long purchase cycles is pointed. Industrial groups, professional associations, standards bodies such as ISO or IEEE, and development finance institutions sit on exactly the kind of primary, authoritative content that AI Mode tends to cite: technical standards, sector research, policy guidance, data sets. They are, in many cases, the category authorities the model would naturally reach for. Their disadvantage is that they have been slower than consumer brands to treat organic citation in AI answers as a strategic objective in its own right, distinct from both paid search and conventional SEO.

The brands best positioned in AI Mode right now are not necessarily spending the most. They are the ones whose content was already structured to answer specific, high-intent questions with named authors, precise data, and institutional backing. That profile matches the serious B2B publisher far more than the typical paid-search advertiser.

The measurement problem

Ad impressions in AI Mode are measurable. Citations are not, at least not through any standard analytics stack. Google does not currently expose citation data via its advertising or search console APIs in any form comparable to click and impression reporting. A brand can know its ad appeared; it cannot know whether it was cited, how its content was characterised in the synthesised answer, or which competitor was quoted instead.

This creates a blind spot that is growing in proportion to AI Mode's query share. Senior marketing and communications leaders at large organisations need a parallel measurement discipline: tracking AI citation independently of paid metrics, using prompt-based audits against the queries their audiences are most likely to pose. Without that, budget allocation decisions rest on ad-server data that accounts for only half the user experience now being served.

The sharper brands will stop optimising paid and organic for AI Mode as if they were the same problem. Winning the auction gets you the ad slot. Getting cited in the answer requires something closer to a publishing strategy than a bidding one.