AI mentions drive delayed site visits that standard analytics miss, Similarweb finds

New analysis from Similarweb reports that brands recommended inside AI assistants such as ChatGPT were 2.5 times more likely to receive a website visit within seven days than brands that were not mentioned. The catch for anyone measuring their own traffic: most of that visit did not arrive as a direct click from the AI tool. It arrived days later as a branded search, when the reader typed the brand name into Google after seeing it recommended by the AI.

Why this matters

If you run a website and judge AI’s effect on your traffic by counting referral clicks from ChatGPT or Perplexity, you will conclude AI is sending you almost nothing. Similarweb’s data suggests that conclusion is wrong. The value shows up on a delay and through a different channel, so last-click analytics quietly credit it to “organic search” or “direct” instead. For a remote-work publisher, the practical takeaway is that being cited in AI answers is worth pursuing even when the referral numbers look flat, and that your own dashboard is probably under-counting the effect.

Context

The finding lands in the middle of a wider debate about whether generative AI search is starving publishers of traffic. It offers a measured counter-point: AI recommendations appear to build brand demand that materialises later, rather than replacing the visit outright. The study’s own limits are worth stating plainly. It covers US desktop users across three verticals, and it establishes correlation rather than proven cause, so the size of the effect for a European remote-work audience is not established.

What to watch

Expect more measurement tools to try to attribute “AI-assisted” branded search directly, which would make the effect visible rather than hidden. Until they do, the honest read is that any brand investing in being the source AI engines cite is likely being rewarded in ways their analytics do not yet show.