Improving Source Trust and Onboarding for Google AI Mode Users
Why people try Google AI Mode and go back to blue links, framed as ranked hypotheses
✳The problem they went after
Google AI Mode swaps keyword queries for conversational, intent-driven questions and returns synthesised answers instead of a list of links. It is embedded straight into the Google search interface, so discoverability is not the issue. Research pointed at two friction points limiting adoption and retention, and they show up for AI-aware and AI-unaware users alike.
✳What they actually did
The hypotheses are ranked rather than listed. The top one is that users distrust results because they have little control over which sources the AI references. Queries sometimes surface irrelevant or outdated links, and when the origin cannot be verified people revert to standard web search. The secondary hypothesis is habit: intent-based questions mean unlearning a decade of keyword search, and answer quality shifts noticeably with phrasing, which discourages a second attempt.
✳What you can take from it
Worth stealing
- Hypotheses ranked by priority, not listed flat
- User trust issue traced through to advertiser revenue risk
- Explicit source filters proposed as the trust fix
- Onboarding cues teaching effective query phrasing
Read this if
PMs on AI search or assistant products worried about repeat usage
The transferable bitThe third concern is the one most submissions would miss. Dynamic source selection makes SEO less reliable, which could pull advertiser spend away from Google as businesses lose confidence in it as a discovery channel. Carrying a user-trust problem through to its advertiser consequence is what makes the recommendations land as business arguments.
✳The submission itself
Open the PDF →Reading on a phone? The embedded viewer is cramped — open the PDF in a new tab instead.
