Briefing · AI visibility
Your business was named in an AI answer. Did anyone walk in?
Being named in an AI answer is one event. Winning a customer is another, and four more steps sit between them. A US experiment with Google Search found that people sent into AI Mode clicked through to other websites less often. It counted clicks to publishers, not visits, bookings or sales.
Published
From a mention to a sale
A mention is the first of five steps
Picture a hypothetical owner opening a visibility report: the business was named in an AI answer. That’s real. But when search gives a complete enough answer, nobody has to click on, and the report can look busy while nothing else moves.
Named
The answer mentions your business.
Cited
The answer points to a source linked to you.
Referred
Someone leaves the answer for an external site or page.
Committed
Someone calls, books or orders.
Became a customer
The booking or order is fulfilled and someone actually becomes your customer.
Each step can happen without the next. One visibility score can’t tell you which held and which broke, so it can’t tell you what to fix.
A Google Search experiment
When search answered in AI Mode, fewer people clicked out
In March 2026, researchers at the University of Pennsylvania and Northeastern University ran a preregistered field experiment with US adults using Chrome and Google Search. For seven days each person had Google unchanged, AI Overviews hidden, or every search sent into AI Mode. The figures below are percentage points: the plain gap between two percentages, not a percent of a percent.
Preprint by researchers at Penn and Northeastern, effect of being assigned to AI Mode
18.8
percentage points lower click-through to other websites
For people assigned to AI Mode, compared with ordinary Google Search.
Preprint by researchers at Penn and Northeastern, effect of being assigned to AI Mode
0.92
fewer search sessions per day
For people assigned to AI Mode, compared with Google left unchanged.
Preprint by researchers at Penn and Northeastern, only where hiding worked
8.8
percentage points higher click-through to other websites
Where AI Overviews were actually hidden. Hiding stopped working reliably, so only 51.1% of them were hidden.
People in AI Mode also rated their searches lower on trust, usefulness, satisfaction, sense of control and relevance. Hiding AI Overviews made no detectable improvement to those ratings.
Why the two click figures aren’t mirror images
In AI Mode, 94.7% of searches were successfully redirected, so the 18.8 figure counts everyone assigned to that condition (researchers call this an intent-to-treat estimate). Hiding AI Overviews worked for about 90% of them on the first day, until Google changed the page code and the browser extension stopped hiding them reliably. So the 8.8 figure counts only where hiding actually worked (a local average treatment effect).
The two figures are not two halves of one clean test, and they are not a plus and minus from one comparison. Both are movements in percentage points, not percentage changes, and both are measured on searches. Nothing here was measured on customers.
The groups differ too. 1,444 people enrolled, completed the first survey and installed the extension, and all spent three days on unchanged Google first, as a before picture. 1,100 searched at least once during the seven days; every behaviour result rests on them. 956 completed the closing survey; every reported opinion rests on them. Neither number counts local-business customers.
Preregistered means the main questions and the analysis plan were published before the confirming results were examined.
Where the evidence stops
It counted clicks to publishers, not customers
Counted
- Clicks leaving Google for publishers and other websites
- Search sessions per day
- How people rated trust, usefulness, satisfaction, sense of control and relevance
Not counted
- Physical locations and local searches
- Google Maps, menus, bookings, orders, purchases or revenue
- Anyone becoming a customer
- Habits over months: each set-up ran for seven days
- Ads in AI Mode: there were no ads during the study
It also doesn’t show that correcting your business information earns a mention, a citation, a click, an order or revenue. Accurate hours and prices are worth having for their own sake.
Who took part, and other limits
Everyone was a US adult using Chrome. The sample skewed younger, more educated and left-leaning.
The paper is a public preprint and has not been peer reviewed, and the raw browsing data are not released.
What to do
Give each step its own proof
If a report only tells you that you were named, that’s what you know. Everything after it needs its own number.
Write down every visibility check
Note the AI system, the app or site you used, the exact question, language, place and date, plus the answer and the source it cited.
Read visits in your own analytics
Don’t infer a visit from a mention. Your own analytics show who actually arrived.
Count calls, bookings and orders where they happen
Your phone log, booking system and till record commitments; a visibility report doesn’t.
Count customers, not intentions
A booking that was kept or an order that was collected is the number that pays the bills.
Keep your facts current anyway
Keep hours, services, prices, availability and policies right for the people who rely on them, without treating that as proof that attention followed.
Keep reading
Related Briefings and next steps
Three related Briefings, and where to start with your own business.
- Most places were never named by AIBeing named, being named first and being described correctly are separate problems. A seven-day study of two Bali submarkets measures each.Read the Briefing
- Google says its AI Search is launching in France. That doesn’t mean it shows your business.Eligibility, appearance, citation, recommendation and a customer’s action are separate states. Check each one.Read the Signal
- AI assistants are moving from recommending to bookingThe wider shift: AI surfaces are starting to act for a customer, not just answer.Read the Briefing
- Check your AI PresenceSee what AI answers say about your business today.See the check
- AI Visibility for local businessesHow a local business gets found and described correctly by AI.See how it works
Sources
Where this comes from
One public preprint, not yet peer reviewed. Obenan has no partnership, endorsement, integration, pilot or special-access relationship with the authors, their institutions, arXiv or Google.
- Full arXiv HTML paper
arXiv Checked
Methods, results, how well each set-up worked, limitations and disclosures.