The practical brief
The finding
Assignment to Google AI Mode reduced external click-through by 18.8 percentage points compared with unchanged Google Search. [c1] [c2] [c4]
Why it matters
An answer can mention your business while the reader stays in search. A visibility report alone cannot tell you whether that attention became a visit, an enquiry or a sale.
What to try
BLURSOR’s practical interpretation
Put answer mentions, search referrals and meaningful website actions alongside one another in your reporting. Ask what each measure tells you before calling a campaign successful.
Study boundary
This was forced use for seven days, not voluntary adoption. The result is not a forecast that your website will lose 18.8% of its traffic.
The result is about clicks per search
The behavioral analysis included 1,100 US Google users. The comparison group kept normal Google Search, where AI Overviews and optional AI Mode were available.
- AI Mode assignment: 18.8 percentage points lower external click-through; 95% confidence interval, 15.3 to 22.2 points lower.
- Click-through was calculated per user as external Google-originating clicks divided by searches. AI Mode conversation turns counted as searches; Google vertical-search pages, such as images, news and shopping, were excluded.
Give each number a job in your report
Use mentions to track whether you appeared, referrals to track arrivals, and your chosen business outcome to track what visitors did. A movement in one column is a question to investigate, not proof of a movement in the next.
Start with the landing pages and customer questions that matter to your business. Keep the date range and question set consistent. If mentions rise while enquiries stay flat, look at the referring visits and landing-page experience before commissioning more content.
- Record the tool, question and date behind each sampled answer.
- Keep referral visits separate from completed purchases, bookings or qualified enquiries.
- Treat this as BLURSOR’s proposed reporting approach; the experiment did not test it.
The news comparison was inconclusive
The separate news-response comparison found no statistically significant change in preference or trust. It does not establish a news-specific trust decline.
Evidence [c3]
The source and its limits
AI in Search Reduces Publisher Referrals Without Improving User Experience: Experimental Evidence
A randomized seven-day Google Search intervention; the highlighted result is the effect of assignment to AI Mode.
- Forced short-term exposure cannot establish long-term voluntary use or a particular website’s traffic loss.
- The US convenience sample was younger and more highly educated than the general population.
Evidence behind this briefing
[c1] External click-through fell by 18.8 percentage points; 95% CI −22.2 to −15.3.
Section 3, Click-Through Rate (H2); Figure 1 and Table S1 · Read the study
[c2] Participants received their assigned search condition for seven days.
Section 2, Experimental Methods and Data · Read the study
[c3] The news-specific preference and trust effects were not statistically significant.
Section 3, News Context (H4) · Read the study
[c4] Click-through uses each participant’s external clicks divided by searches.
Supplement S2.4, Primary Outcome Measures: Click-Through Rate · Read the study