Similarweb finds AI-influenced visits arrive through search more often than other visits

Similarweb reports that 55.9% of AI-influenced visits arrived through search, against 40.4% of visits without AI influence. The figures were published in a Neil Patel write-up of the Similarweb research.

What Similarweb measured

Similarweb followed users who asked ChatGPT an industry-related question and received a specific brand recommendation. It then tracked their behaviour over the following seven days.

• Across finance, travel and beauty, users who received a recommendation for a brand were 2.5x more likely to visit that brand's website than users who received a recommendation for a competing brand.

• The analysis covered users who had not previously visited the recommended brand and had not named it in their prompt.

• AI-influenced visitors viewed nearly twice as many pages and spent roughly twice as much time on site as other visitors, according to the same write-up.

The search figure

The 55.9% is the share of AI-influenced visits that arrived through search. The 40.4% is the same share for visits without AI influence. A visitor who sees a brand in a ChatGPT answer and later searches for it on Google will most likely be logged as organic search. The AI step gets no credit.

The write-up does not state a sample size, and we have not seen the Similarweb report itself. Treat the numbers as Similarweb's, as relayed by Neil Patel.

What the number leaves open

Hidden AI traffic is usually assumed to sit in Direct, because AI referrals often lose their referrer. This data adds a second place to look. It also leaves several questions unanswered:

• The write-up says "search". It does not split that into branded and non-branded queries.

• It does not give the share of AI-influenced visits that arrived as Direct, so the two leaks cannot be compared from this source.

• Similarweb shows an association between AI recommendations and later visits. The source states that this does not prove every later visit was caused by the recommendation.

What to do with it

Last-click reports will undercount AI if the AI step happens before a branded search. The practical response is to read several signals together:

• Track branded search demand alongside your visibility in AI answers.

• Check whether branded query growth moves with gains in AI visibility.

• Look at direct visits, engagement and downstream conversions next to AI referral traffic.

None of these proves causation on its own. Together they show relationships that referral reporting cannot see.