The traffic model assumed a click. A buyer searched, chose a link, and arrived with a referrer attached. When the answer appears in the results, that chain breaks — and everything downstream of it stops measuring what you think it measures.
The scale, honestly stated
Per the Reuters Institute's 2026 trends report, Google organic referrals to more than 2,500 news sites fell 33% globally and 38% in the US between November 2024 and 2025, while AI Overviews appear in roughly 10% of US search results. The same report notes Google still delivers around 500 times more referrals than ChatGPT from search alone. Both facts are true: traditional search remains the volume, and its referral share is eroding.
Where the effect actually surfaces
- Branded search — buyers who saw you in an answer later search your name directly.
- Direct traffic — the same effect, arriving without a referrer at all.
- Assisted conversions — a longer path where the first touch is invisible.
- Share of answer — the leading indicator that moves before any of the above.
Why the conventional response is wrong
Two reactions are common and both fail. The first is abandoning SEO because clicks are falling, which discards the surface still delivering most referrals. The second is ignoring AI surfaces because they do not show in analytics, which is optimising for what is measurable rather than what matters.
The correct response is to change what you measure. Track share of answer alongside sessions, watch branded search as a leading signal, and state plainly which numbers are causal and which are correlational.
What this means for your business
If your reporting is session-based only, you are now partially blind. Add share of answer and branded search to the dashboard before concluding anything about performance. Our comparison of GEO and SEO covers how the two surfaces differ.