Evidence → Queue

Lesson 7 · The product

The queue: 53 clusters, 17 survivors, zero padding

Every candidate is a PRODUCT_HYPOTHESIS with a persona, verbatim receipts, explicit unknowns, kill criteria and a cheapest next test. 36 clusters were rejected with named reasons — that is what zero padding looks like.

0

DO_NOW

No observed multi-package WTP anywhere in the corpus. The queue says so, out loud.

17

WATCH

Persona + pain + falsifiable next test. Run the test; the queue never says build.

36

Rejected

single_package_recurrence (15) · no_icp_detected (11) · no_theme_match (6) · no_pain_evidence (4)

Three survivors, as examples

CandidatePersonaEvidenceCheapest next test
agentic-codingdevelopers / engineering teams8 pain signals · 4 packages · both families5 problem-interviews + landing page; success = ≥5/20 replies with a live problem statement
local-servicesprofessional services (legal/dental/accounting)5 pain · 5 packages · both familiesICP interviews; kill = pain confirmed but unpaid
customer-support-bposupport / BPO operations5 pain · 4 packages20 direct outreaches; failure = ≤1 reply

What every candidate carries

persona_icp · pain counts with package spread · WTP status (honest NONE when absent) · hypothesis with falsifier · assumptions · kill criteria · proposed wedge · why_now (or UNKNOWN) · best 3 verbatim quotes with locators · explicit conflicts · unknowns and missing evidence · transparent score breakdown with formula_version · a next_test with question, action, effort, owner, metric, success and failure thresholds.

"A queue that says DO_NOW without observed payment is a horoscope. This one says WATCH and hands you the cheapest experiment."Design principle, queue-scoring-v2
Lesson 07 of 8 · Evidence → Queue