Google's AI answer sharing plans in Search: what changes for SEO attribution

On 23 October, the French SEO publication Abondance reported on a Google Careers listing for a Senior Product Manager, AI Google Search Growth, based in Mountain View. The role owns the roadmap for the "Share System" and for "Live Notifications" across Search, and is described as "0 to 1" work: building sharing mechanics for AI answers that do not exist yet. The posting asks for experience with share growth funnels, experimentation and generative AI interfaces. It does not say what Live Notifications will be.

A job listing signals a priority, not a launch date. It does, however, extend a feature that already exists. Since summer 2025 in the US, users can create a link from an AI Overview or an AI Mode answer and send it through any messaging app. The recipient opens it in AI Mode and can keep asking questions. They land on Google, not on a publisher's page.

The shared unit moves from the link to the answer

For two decades, sharing a search result meant sharing a URL, and that URL sent a visit to whoever published the page. The analysis relayed by Abondance, from SEO consultant Eli Schwartz, argues that this is ending. When people share the AI answer itself, the cited site loses both the link and the visit. He calls the result "AI Social": the merger of dark social (private shares that show up as direct traffic, if at all) and influence that produces no measurable click.

His scenario, which is his own reading and not part of the posting, pairs sharing with notifications: two people plan a trip or watch a price in AI Mode, and both get alerted when something changes. Search would then keep users coming back without a new query. If that happens, the conversation continues inside Google, and a brand that was missing from the first answer has little chance of entering it later.

What remains unknown matters as much: the scope of Live Notifications, whether shared answers will point more visibly to cited sources, and whether any of this will launch outside the US.

What changes for acquisition and measurement

For teams that depend on organic search, the practical consequence is attribution decay more than a sudden traffic drop. Part of the influence your content has will reach people who never click, and part of the visits that do arrive will be logged as direct traffic. Last-click models will undercount search even more than they already do.

On content, the conclusion matches what many SEO teams already see in their data: top-of-funnel informational pages (definitions, generic guides, short answers) keep losing referral clicks. Visibility on more qualified, mid-funnel intents, where a user is comparing options or about to decide, becomes the part worth fighting for.

What to check this week

• Audit whether you are cited in AI Overviews and AI Mode for your most valuable commercial queries, not only your highest-volume ones.
• List the follow-up questions a user, or the person they share the answer with, would ask next, and check that you have a page answering each one.
• Compare the trend of direct traffic with branded search volume: both rising while organic clicks stay flat suggests that unattributed discovery is growing.
• Add a "how did you hear about us" field to sign-up or checkout if you do not have one; self-reported attribution captures what analytics cannot.
• Review which informational pages exist only to collect traffic, and decide whether they still justify their maintenance cost.

None of this requires waiting for Google to ship anything new. The sharing feature is already live in the US, and the measurement gap it opens is already visible in analytics for anyone who looks at direct traffic and branded demand side by side.