Probative critique: checkout-prd.md

test/model · 1 document · 431 characters · 0.0s · 10 successful model calls

2 blocking · 1 warnings · 0 notes

Summary

One row per critic. Clean is the share of candidates checked that drew no blocking or warning finding; the floor is the severity veto.
Dimension	Checked	Clean	Block	Warn	Floor (0 to 10)	Status
Evidence	2	1 of 2 (50%)	1	0	0.0	BLOCKED · small sample
Problem framing	2	1 of 2 (50%)	1	0	0.0	BLOCKED · small sample
Segments	1	1 of 1 (100%)	0	0	10.0	clean · small sample
Story quality	1	1 of 1 (100%)	0	0	10.0	clean · small sample
Dependencies	0	—	0	0	—	nothing found to check
Constraints	0	—	0	0	—	nothing found to check
Stress test: least sure about	3	2 of 3 (67%)	0	1	8.0	warnings · small sample

Findings

Blocking

Evidence: The statement presents a specific figure, proportion, ranking, benchmark or study finding as fact, and gives no basis for it. (1 of 2)

✗ BLOCK “FLAG Competitors convert far better than we do.” checkout-prd.md, line 4

# Checkout PRD CLAIM: Conversion drops 12% at the payment step. CLAIM: FLAG Competitors convert far better than we do. NEED: Users need a FLAG <script>alert(1)</script> | `tick` **bold** [x](http://evil.test) NEED: Users want to pay without retyping details. STORY: As a shopper I want to save my card so that I can pay faster. SEGMENT: Adults aged 25 to 45 …

Name where the figure comes from, or restate it as an assumption to be tested

Problem framing: The need names a feature, a screen or UI element, an implementation, or a named technology instead of the customer benefit. (1 of 2)

✗ BLOCK “Users need a FLAG <script>alert(1)</script> | `tick` **bold** [x](http://evil.test)” checkout-prd.md, line 5

Restate as a customer benefit — verb first, customer voice

Warnings

Stress test: least sure about: The statement reports a change, trend, pattern or correlation, links it in its own words to a cause or a conclusion (because, so, driving, shows that, clearly, which is why, proves), and rules out no other explanation: seasonality, a change in who or what was counted, a concurrent release or event, a selection effect. (1 of 3)

▲ WARN “FLAG Competitors convert far better than we do.” checkout-prd.md, line 4

Would have to be true: For “FLAG Competitors convert far better than we do.” to support the conclusion drawn from it, the other plausible explanations must have been excluded.

Name the other explanations and how each was excluded, or state the conclusion as a hypothesis

Not assessed

Dependencies: nothing found to check
Constraints: nothing found to check

Coverage

Document	Format	claim	constraint	dependency	forecast	need	segment	story
checkout-prd.md	markdown (detected)	2	0	0	1	2	1	1

100 input tokens · 50 output tokens · 1 concurrent calls · critics reply with flagged items only · probative 0.0.0-test