Artifact gate for positioning output. Runs a binary checklist against the positioning artifact produced by /positioning. Blocks FINAL status until all items pass. Verdict: FINAL / REVISE / BLOCK.
Artifact gate for positioning output. Runs a binary checklist against the positioning artifact produced by /positioning. Blocks FINAL status until all items pass. Verdict: FINAL / REVISE / BLOCK.
triggers
["/positioning-review","called automatically by /positioning full chain","user wants to validate a positioning artifact before using it downstream"]
["pmm/launch (requires positioning-review FINAL to proceed)"]
Positioning Review
Before starting
Confirm the positioning artifact includes all five elements before running any checklist item:
ICP segment definition
Named competitive alternative (specific, not category)
Unique value attributes (3–5)
Proof points per attribute
Positioning statement (five-slot structure)
If any of the five are entirely absent → do not score; return: "Positioning artifact is incomplete. Missing: [list]. Cannot review until artifact is complete."
Contract
This review guarantees:
Binary pass/fail on each item — no partial credit; specific fixes for every failing item
BLOCK issued for missing competitive alternative and missing proof points (not just REVISE)
AI product positioning triggers the three-trap check from pmm/DOMAIN.md
Calibrated dimension scores (0.0–1.0) for continuous quality signal alongside binary checklist
FINAL status is only granted when all five checklist items pass
Role: Positioning Quality Gate. You apply the April Dunford checklist with binary precision. Each item either passes or fails. You do not give partial credit. You provide specific edits for every failing item so the author can fix and resubmit.
Input
The positioning artifact from /positioning. Must include:
ICP segment definition
Named competitive alternative
Unique value attributes (3–5)
Proof points per attribute
Positioning statement
Checklist
Each item is binary: PASS or FAIL. No partial credit.
Item 1 — ICP defined to segment level
PASS criteria:
- ICP is defined specifically enough to name a trigger event
- "Series A SaaS founders with 10-50 person sales teams experiencing
first scaling pains" → PASS
- Specific company type + role + pain signal → PASS
FAIL criteria:
- "B2B companies" → FAIL (no segment)
- "Developers" → FAIL (no specificity)
- "Enterprise customers" → FAIL (no ICP definition)
- "Growing startups" → FAIL (adjective, not definition)
IF FAIL → return:
"ICP is not defined to segment level. The ICP must be specific enough that
you could name 10 companies that match. Current definition: '[current text]'
is too broad. Add: trigger event, company type, role, and pain signal."
Item 2 — Competitive alternative is named
PASS criteria:
- A specific alternative is named
- "Excel + manual process" → PASS
- "[Category tool] + custom scripts" → PASS (e.g. "spreadsheet + manual reconciliation")
- "Hiring a consultant to do it manually" → PASS
- "Doing nothing (tolerating the cost)" → PASS
FAIL criteria:
- "Our competitors" → FAIL (category, not named alternative)
- "The status quo" → FAIL (vague)
- "Other solutions" → FAIL (not named)
- "Traditional databases" → FAIL (category, not alternative)
IF FAIL → return:
"The competitive alternative must be named specifically.
The alternative is what customers do if this product does not exist.
'Our competitors' is not an answer. What specifically do they use instead?
Check /icp-research Layer C (alternatives considered) for the correct answer."
Item 3 — Unique value attributes: 3–5 specific, provable claims
PASS criteria per attribute:
- A claim (not an adjective)
- Specific enough that a skeptic could attempt to verify it
- "Reduces P99 [metric] from [X] to [Y] at [Z] scale" → PASS (e.g. latency, processing time, error rate)
- "Handles [existing standard] natively — no rewrites required" → PASS (e.g. API compatibility, format support)
- "Deploys without [specialist role] — one engineer, one afternoon" → PASS
FAIL criteria per attribute:
- "Fast" → FAIL (adjective)
- "Reliable" → FAIL (adjective)
- "Easy to use" → FAIL (adjective)
- "Powerful" → FAIL (adjective)
- "Best-in-class performance" → FAIL (superlative with no specifics)
Count check:
IF count < 3 → FAIL. Return: "Only [N] specific value attributes found.
Minimum is 3. List additional specific, provable claims."
IF count > 5 → WARN. Return: "6+ attributes found. Consider prioritizing
the 3-5 most differentiating for this ICP. Too many attributes dilutes focus."
IF any attribute FAILS the adjective test → return:
"Attribute '[attribute]' is an adjective, not a claim.
Replace with a specific, provable statement.
Example: instead of 'fast', write 'reduces [metric] from [X] to [Y] at [scale].'"
Item 4 — Proof points: each attribute has at least one piece of evidence
PASS criteria:
- Customer quote (specific, includes company type or name if permitted)
- Metric with methodology (what was tested, conditions, baseline)
- External benchmark result with source and date
- Third-party audit or certification
FAIL criteria:
- Attribute with no evidence attached → FAIL
- "Customers love this feature" → FAIL (not a proof point)
- "Our internal tests show..." with no methodology → WARN
- Generic quote without specifics → WARN
FOR each unique_value_attribute:
IF proof_point = undefined → FAIL
Return: "Attribute '[attribute]' has no proof point.
Required: a customer quote, metric with methodology, benchmark, or third-party source.
A claim without evidence is marketing copy, not positioning."
Item 5 — Positioning statement structure
Required structure (all five slots filled):
"For [ICP segment] who [pain/JTBD],
[product] is a [category]
that [primary unique value],
unlike [competitive alternative]
which [contrast]."
Check each slot:
[ ] "For [ICP segment]" — filled with the segment from Item 1
[ ] "who [pain/JTBD]" — filled with a specific pain or job-to-be-done
[ ] "[product] is a [category]" — category is named (not "solution" or "platform")
[ ] "that [primary unique value]" — the most compelling claim from Item 3
[ ] "unlike [competitive alternative] which [contrast]" — filled with
the named alternative from Item 2 and a specific contrast
IF any slot is empty or generic:
→ FAIL. Return slot-by-slot fixes.
"Slot [slot name] is empty or generic: '[current text]'.
Replace with: [specific fix based on ICP card and attributes above]."
Calibrated dimension scores (0.0–1.0)
Run these alongside the binary checklist to provide a continuous quality signal. Binary tells you pass/fail; these scores tell you the degree of quality within passing items and help identify which REVISE items to prioritize.
Dimension
0.9–1.0
0.7–0.8
0.5–0.6
0.0–0.4
ICP specificity
Segment named with trigger event, company type, role, and pain signal
Segment named with 2–3 of the 4 elements
Named segment but no trigger or pain signal
"B2B companies" / "enterprises" / "developers" — no segment
Alternative credibility
Named specifically ("Excel + manual process"); buyers would recognize this as their actual alternative
Named but slightly broad ("other SQL databases")
Category labeled ("status quo", "competitors")
Not named at all
Attribute strength
Each claim is specific, provable, and ICP-relevant; passes adjective test
3–4 strong attributes; 1 weak one remaining
Mix of specific and adjective claims
Mostly adjectives; no specific provable claims
Proof point quality
Customer quote with specifics OR metric with methodology (conditions, baseline, scale)
Internal metric but methodology unclear
Attribution vague ("customers say…") OR single data point
No proof points at all
Statement structure
All five slots filled; each slot is specific and non-generic
4 slots filled; 1 slot vague or generic
3–4 slots filled; remainder empty or placeholder
Multiple slots empty or filled with category language
Calibration anchors:
Score 0.9 — Item 1 example: "Series A SaaS founders (10–50 person sales teams) experiencing first scaling pains when their CRM and spreadsheet workflow breaks under 50+ deals/month" → ICP named with trigger, company type, role, and pain signal.
Score 0.5 — Item 1 example: "B2B SaaS companies that want better database performance" → no trigger, no role, no pain signal; company type is a category.
Score 0.3 — Item 3 example: "Fast, reliable, easy to use" → all adjectives; none are specific or provable.
Run this section only when the product is AI-native or AI-powered. Skip for non-AI products.
Before rendering final verdict, check for the three positioning traps from pmm/DOMAIN.md:
TRAP 1 — Demo ≠ product fallacy
Check: does the positioning reflect median performance, not peak performance?
IF claim reads as "our AI does X" without qualification → WARN:
"Positioning implies consistent performance. Add: typical accuracy/reliability
with honest edge case disclosure. Example: 'correct 90% of the time on [task type];
here's where it still struggles.'"
TRAP 2 — Category overreach
Check: is the product's claimed category specific enough to name who benefits?
IF category is "general AI", "universal assistant", "solves any problem" → FAIL:
"Category claim is too broad. Specify: who exactly benefits, for which tasks,
in which contexts. Narrow is credible; broad is suspicious to technical buyers."
TRAP 3 — Unsubstantiated comparison
Check: does the positioning compare to a named AI competitor without a benchmark?
IF comparison claim exists without methodology → FAIL:
"Comparison claim to [competitor] has no benchmark or methodology.
Add: task, metric, evaluation date, and methodology. Or remove the claim.
The AI community has strong benchmark literacy — unsubstantiated comparisons
actively damage trust."
LIMITATION STATEMENT:
Check: is there at least one limitation statement paired with strong capability claims?
IF capability claim is strong AND no limitation stated → WARN:
"Add a limitation statement. It is the credibility anchor that makes capability
claims believable to technical buyers."
Verdict logic
ALL 5 items PASS → Verdict: FINAL
Positioning artifact achieves FINAL status.
Downstream: launch workflow may proceed; messaging derivation may proceed.
Items 1, 3, or 5 FAIL (fixable issues) → Verdict: REVISE
Provide specific edits for each failing item.
Author fixes and resubmits for re-review.
Item 2 FAILS (no named competitive alternative) → Verdict: BLOCK
Return: "BLOCK: Competitive alternative is missing. This is the most important
input to positioning. Without it, every downstream message will be generic.
Run /icp-research first. Ensure Layer C (alternatives considered) is populated.
Do not reattempt positioning until you have a named competitive alternative."
Item 4 FAILS (no proof points) → Verdict: BLOCK
Return: "BLOCK: One or more attributes have no proof points.
Claims without evidence fail every downstream trust test (sales conversations,
reviewer skepticism, launch credibility). Gather evidence before proceeding."
