Candidate-set cases

Two products. Two different AI shopping decisions.

Each case begins with a real buyer need, separates retrieval from candidate-set inclusion and comparison, and ends with the smallest useful next decision.

CAPTURED CASES

Read the evidence path—not only the final order.

These cases report captured outcomes under recorded conditions. They do not establish stable rankings or guarantee future inclusion, traffic or sales.

CASE 01 · STORE-REVIEWEDSleep product

The right product. Missing from the shortlist.

A need-led test and material-aware control separated product fit from a weak category connection.

Read case 01 →
CASE 02 · INDEPENDENT RESEARCHUSD 99 Snapshot

Readable by name. Missing by need.

The product passed direct retrieval, then remained absent through one four-turn unbranded shopping conversation.

Read anonymous case 02 →