AI answers are changing how consumers research insurance and mortgages before they ever see a form. What the measured data says, and what it does to your pipeline.
Figures in this article describe the wider market and are drawn from the third-party sources listed at the end. They are not Lead Foundry results, and nothing here is a projection of what any individual broker will achieve.
Consumers researching cover or a mortgage increasingly get their orienting answer without visiting a website. That changes the shape of the enquiry that eventually reaches you, and it changes what a lead is worth.
Pew Research Center ran a behavioural study of 900 US adults covering 68,879 unique Google searches, of which 12,593 produced an AI summary, using browsing data from March 2025 with results captured 7 to 17 April 2025.
When an AI summary was present, users clicked a traditional result on 8% of visits. Without one, 15%. Users clicked a link inside the AI summary itself on just 1% of visits.
| Clicked a result, no AI summary present | 15% |
|---|---|
| Clicked a result, AI summary present | 8% |
| Clicked a link inside the AI summary | 1% |
The middle bar is roughly half the first. The third is the one that matters for anyone hoping citation traffic replaces click traffic.
Source: Pew Research Center
The broader trend is consistent. Search Engine Land reported that US Google searches ended without a click 68.01% of the time across the first four months of 2026, up from 60.45% in 2024. Similarweb data puts Google AI Mode referral rates at roughly 1.6% to 2.5% of queries against a traditional Google referral rate of 17% to 19%.
Three effects, and they do not all run the same direction.
| Effect | What changes | Whether it helps you |
|---|---|---|
| Fewer, later-stage enquiries | The comparison happens before the form | Helps, if you can read intent |
| Less traceable history | You cannot see what shaped their view | Hurts, and makes the first call harder |
| Consumers arrive with an anchor | They have been told something already | Neutral, if you ask before correcting |
| Attribution degrades | Self-generated funnels get harder to measure | Hurts anyone building their own supply |
The first-call consequence: A consumer who has been given an answer by a chatbot has an anchor, and the anchor may be wrong. Ask what they have been told before correcting it. Opening by contradicting an unstated belief is the fastest way to lose a call you had already paid for.
Because being the source an answer engine draws on is now a distribution position. A site that answers a question clearly, carries checkable citations, and is structured so passages can be lifted is more likely to be the source of the summary a consumer reads.
An operator who owns their consumer brand can build that. An operator buying traffic from affiliate networks is downstream of whoever did, and is competing for a shrinking pool of clicks rather than for the answer itself.
It reduces clicks and compresses the research stage. Pew Research found users click a traditional result on 8% of visits when an AI summary is present against 15% without one, and click a link inside the summary on about 1% of visits. US Google searches ended without a click 68.01% of the time in early 2026, up from 60.45% in 2024.
The measured data says yes. Pew Research's behavioural study of 68,879 searches found the click rate on traditional results roughly halves when an AI summary appears. Similarweb puts Google AI Mode referral rates at roughly 1.6% to 2.5% of queries against a traditional referral rate of 17% to 19%.
Enquiries arrive fewer in number, further along the decision, and with far less traceable history. That raises the value of records carrying real intent signals, and it raises the value of suppliers who own their consumer brands, since being the source an answer engine cites is now a distribution position rather than a byproduct of ranking.
Ask what the consumer already believes before correcting anything. People arriving from an AI summary carry an anchor that is invisible on the lead record and may be wrong. Opening by contradicting an unstated belief loses calls that were otherwise winnable.