Study: Google AI Mode sends fewer clicks, worse experience

Google's own line on AI Mode has been that it sends "higher quality" clicks, users who stay longer and engage more once they land on a site. Search Engine Journal reports that new research contradicts this, finding AI Mode reduces the number of clicks a page receives and delivers a worse user experience than traditional search results. That is not a nuance. It is a direct rebuttal of the company's chosen defence of a product that is already reshaping how enterprises get found online.

The distinction matters because "quality over quantity" has been Google's answer to every publisher complaint about traffic collapse since AI Overviews launched. Fewer clicks, the argument goes, but better ones: visitors who convert, read, subscribe. If that trade-off were real, it would be a survivable story for brands. Less traffic but more valuable traffic is a shape publishers have adapted to before. What the new research suggests is that there is no trade-off. There is just less traffic, of no better quality, arrived at through a worse path.

Why the click was never the point

The deeper issue is that AI Mode was never built to send clicks. It was built to answer the question inside Google's own interface, using content scraped, summarised and re-served without a visit to the source. A click was always going to be the residue, not the objective. Framing that residue as "quality" was a marketing choice, not a description of the underlying architecture. The research reported by Search Engine Journal simply removes the cover.

For the sectors that matter to this publication, the implication runs in a specific direction. Financial services firms, multilateral bodies and industrial groups have spent two years optimising content for citation inside AI Overviews and AI Mode, often at the direct expense of the classic ten blue links that used to reward well-structured, authoritative pages with a click. If that optimisation was justified by the promise of denser, higher-value traffic downstream, this research pulls the justification out from under it. The honest accounting is: you are being cited, and you are not being visited, and the citation is not compensating.

That has a concrete consequence for how these organisations should measure success. A think tank tracking its citation frequency in AI Mode without pairing it against referral traffic and session quality is measuring half the picture, and the flattering half at that. A bank's investor relations team that treats a Gemini citation of its ESG report as equivalent to a website visit is making a category error Google itself, per this research, cannot substantiate. The metric that matters is not "are we cited" but "does the citation do anything for us once it appears." Increasingly the answer looks like no.

There is a second, quieter cost. A poor user experience inside AI Mode does not stay contained to Google's interface. If users find AI Mode's answers thinner, slower to trust, or harder to verify than a direct source, that friction becomes an argument for going elsewhere: to Perplexity, to ChatGPT with browsing, to a direct site search. Google's dominance in search has rested on being the default, not necessarily the best, for two decades. Research that shows its flagship AI feature underperforms the format it is replacing gives every rival a talking point and every brand a reason to hedge its AI-visibility strategy across more than one model.

None of this argues for retreating from AI-mode optimisation. Citation inside Google's AI surfaces remains a real channel of influence, particularly for institutions whose authority depends on being the source a model reaches for. But the research reported by Search Engine Journal is a useful corrective to a specific complacency: the idea that being cited is a proxy for being visited, and that Google's assurances about traffic quality can substitute for a brand's own measurement. Enterprises that have not yet built independent tracking of what an AI Mode citation actually delivers, in clicks, in time on site, in conversion, are relying on the word of a company whose own claim has just been contradicted by outside research. That is not a comfortable position for a CMO to defend to a board.