Only 1.1% of news publisher visits come directly from AI: Study

AI referrals accounted for conscionable 1.1% of visits to news publishers aft conversations pinch AI assistants, according to a caller Scrunch study that tracked millions of searches, AI chats, and website visits.

Most patient visits came done nonstop navigation aliases accepted search. Google AI Overviews mostly appeared connected news searches that already generated comparatively fewer patient clicks.

By the numbers. Readers were 20.5 percent points much apt to sojourn a news patient during the week aft a news-related AI speech than aft a non-news conversation. The study compared each reader’s browsing history. Scrunch said the results do not show that AI created caller demand, only that news-related AI conversations often coincided pinch continued news reading.

• Among those post-chat visits, astir three-quarters came done nonstop navigation.
• About 9% came from accepted search, and 1.1% carried an AI referrer.
• Scrunch noted that referral information captures only the past click earlier a visit. It can’t reconstruct the afloat way crossed AI conversations, searches, and later website visits.

AI Overviews appeared connected lower-click news searches. News searches pinch an AI Overview generated patient clicks astir 20% of the time, compared pinch astir 30% for searches without one.

• When Scrunch compared the aforesaid query pinch and without an AI Overview, the spread shrank to astir 2 percent points.
• The study recovered that AI Overviews astir often appeared connected news searches that already produced comparatively fewer patient clicks.
• During the study period, AI Overviews appeared connected astir 1 successful 4 news searches. They were astir communal for utility-style topics specified arsenic weather, marketplace prices, and explainers.
• Sports searches, which had the highest patient click-through rates, seldom triggered an AI Overview.

Large publishers captured astir follow-through visits. When an AI adjunct mentioned a patient that the scholar had not included successful the prompt, visits to that patient accrued by 10.6 percent points the adjacent day. They accrued by astir 20 percent points complete the pursuing week, compared pinch akin publishers not mentioned successful the aforesaid response. Scrunch described this arsenic a routing signal, not grounds of causation.

• Major publishers accounted for 82% of patient names volunteered by AI assistants and 97% of follow-through visits.
• Mid-sized publishers made up astir 17% of patient mentions but received a overmuch smaller stock of resulting visits.

Why we care. If you measurement AI’s effect only done referral traffic, you are apt missing interactions that later lead to nonstop visits aliases accepted searches. The study suggests search your visibility successful AI answers alongside referral information to amended understand really group observe your brand.

About the data. The study analyzed millions of searches, AI conversations, and patient visits from a privacy-safe, opt-in sheet betwixt February and June 2026. It measured page visits, not subscriptions, revenue, aliases clip spent.

The study. Surviving the clickpocalypse: What millions of hunt events opportunity astir AI eating the news