| name | content-review |
| description | Scores content pieces and experiment specs before publication or launch. Part A: content scoring (8 dimensions, 0.0–1.0 each). Part B: experiment spec scoring (8 dimensions). Calibrated examples anchor scores across sessions. Verdict: APPROVE / REVISE / REJECT. |
| triggers | ["/content-quality-review","user wants to review a piece of content before publishing","user wants to review an experiment spec before running it","called by launch workflow on primary launch copy"] |
| role | review |
| type | quality |
| version | 1.1.0 |
| sources | ["b2b-practitioner-voice (growth-skills v1.0, score 9/10)","growth-quality-rubric (growth-skills v1.0, score 9/10)","growth-publish-safety (growth-skills v1.0, score 6.5/10)"] |
Content Quality Review
Before starting
Confirm (ask or infer):
- Content type — Part A (content piece: blog, email, landing page, social) or Part B (experiment spec)?
- Destination platform — HN / LinkedIn / X / Email / Product Hunt / landing page? (determines platform format dimension)
- Intended audience / ICP — if available, improves ICP alignment scoring; if not available, note in output
- Brain context — is
playbooks/messaging.md available? If yes, check approved messaging before scoring brand voice and differentiation
- AI product — if yes, apply AI-specific banned phrases from
pmm/DOMAIN.md in addition to standard list
Contract
This review guarantees:
- Banned hype pattern scan runs before any scoring — patterns found trigger mandatory rewrite note
- All 8 dimensions are scored (0.0–1.0) before verdict is rendered
- APPROVE requires average ≥ 0.75 AND zero banned patterns
- Specific fixes are provided for every dimension scoring < 0.7
- Verdict logic is applied consistently using calibrated examples as anchors
Role: Quality Auditor. You score against the practitioner voice standard and a calibrated rubric. You do not give vague notes like "make it more compelling." You give specific fixes for every dimension that fails. You use calibrated examples to keep scores consistent across sessions.
Inputs
Required before proceeding:
- The artifact to review: one of the following:
- Part A (content): A piece of written content — blog post, landing page copy, email, ad copy, social post, or similar
- Part B (experiment spec): A growth experiment specification — hypothesis, metric, channel, distribution plan
- The intended audience or ICP context (if available — improves scoring accuracy for dimensions like Specificity and Relevance, but not required to proceed)
If no artifact is provided:
NEEDS_CONTEXT. Return:
"Please provide the artifact to review. This can be:
- A piece of content (copy block, email, blog post, landing page text)
- An experiment spec (hypothesis, metric, channel, timeline)
Paste the full text or share a link."
Banned hype patterns
Before scoring, scan for banned patterns. Any of these in non-quoted text triggers a mandatory rewrite note, regardless of other scores:
| Banned phrase or pattern | Replacement direction |
|---|
| "game-changing" | Describe the specific change it causes |
| "revolutionary" / "revolutionizing" | Describe what is different now vs before |
| "unlock" (as verb for potential/capability) | Use concrete: "enables", "lets you", "removes the need to" |
| "leverage" (as verb for taking advantage) | Use: "use", "apply", "build on" |
| "empower" | Describe the specific action enabled |
| "seamless" | Describe what friction was removed |
| "cutting-edge" / "state of the art" (without benchmark) | Add benchmark or remove claim |
| "robust" (vague) | Describe the specific reliability property |
| "world-class" / "best-in-class" | Add specific proof or remove |
| "innovative" / "transformative" / "paradigm" | Describe the mechanism |
| "unprecedented" | Describe why it is new, specifically |
| "excited to announce" and similar hollow enthusiasm openers | Start with the news, not the emotion |
| "powerful AI" | Describe the specific capability |
| "proprietary model" (without specifics) | Describe what makes it different |
| "state of the art" (without benchmark) | Add benchmark with date and methodology |
| "hallucination-free" | Replace with accuracy metric with methodology |
| "easy to use" (without specifics) | Describe what friction was removed |
| "transforming" / "reimagining" / "redefining" | Describe the specific change |
| "just dropped" / "someone just built" (without specifics) | Name the specific thing, who built it, and why it matters — vague hype opener |
| "insane" / "wild" / "absolutely insane" (as quality descriptors) | Replace with the specific number, behavior, or outcome that is surprising |
| "everyone needs to see this" / "you need to know about this" | State what the thing is and why it matters to this specific reader |
| "the future of [category]" (without evidence) | Describe the specific mechanism or shift, not the future state |
| "things will never be the same" |
Rule: banned patterns found in a quote from a third party are fine. Banned patterns in the author's own voice fail.
Part A — Content piece scoring
Use for: blog posts, LinkedIn posts, HN posts, emails, landing page copy, case studies, social content.
Dimensions (score each 0.0–1.0)
Dimension 1 — ICP alignment (0.0–1.0)
Does this content speak directly to the target ICP's pain, context, and language?
| Score | Description |
|---|
| 0.9–1.0 | Names the specific ICP pain; uses their vocabulary; they would recognize this as written for them |
| 0.7–0.8 | Clearly relevant to the ICP; slight mismatch in language or pain specificity |
| 0.5–0.6 | Generally relevant to the category; could apply to many audiences |
| 0.3–0.4 | Weak ICP signal; mostly product-centric, not customer-centric |
| 0.0–0.2 | No ICP alignment; generic or wrong audience entirely |
Dimension 2 — Funnel stage fit (0.0–1.0)
Does the content depth and CTA match the appropriate awareness/consideration/decision stage?
| Score | Description |
|---|
| 0.9–1.0 | Depth is right for the stage; CTA asks for appropriate commitment |
| 0.7–0.8 | Mostly right; small mismatch (e.g. awareness content with hard CTA) |
| 0.5–0.6 | Stage is ambiguous; content tries to do multiple stages |
| 0.0–0.4 | Wrong stage entirely (e.g. decision content sent to cold audience) |
Dimension 3 — CTA appropriateness (0.0–1.0)
Is the CTA approach correct for the content type and audience? Is the one next step honest about effort required?
| Score | Description |
|---|
| 0.9–1.0 | CTA matches channel norms; one clear next step; honest ("takes 5 minutes" not "instant") |
| 0.7–0.8 | Mostly right; effort slightly misrepresented or secondary CTAs compete |
| 0.5–0.6 | CTA present but approach mismatches the channel (e.g. explicit on HN, absent on campaign page) |
| 0.0–0.4 | Wrong CTA approach for the audience; multiple conflicting CTAs; deceptive framing |
CTA calibration by channel:
| Channel | Appropriate CTA | Flag |
|---|
| HN / Show HN post | None, or "questions welcome" | Any conversion language — gets downvoted, damages credibility |
| GitHub README | Link to docs; "try it" with zero friction | Pricing language, signup gate |
| Developer Discord / Slack | Soft invitation to explore | Upgrade asks, pricing language |
| Technical blog / case study | Low-friction: "try for free," "read the docs" | Demo request forms embedded in editorial body |
| LinkedIn (practitioner post) | Explicit CTA acceptable | Multiple competing CTAs in one post |
| Reddit (engineering subs) | Soft: link to full article | FOMO language, "sign up now" |
| Campaign landing page | Explicit CTA required | No CTA, or CTA buried below multiple feature sections |
Progressive CTA sequence when explicit CTA is appropriate: free/zero-commitment first → engagement/community second → high-commitment ask last. Never lead with the highest-commitment ask.
Dimension 4 — Platform format (0.0–1.0)
Does the content fit the native norms of the platform it's going to?
| Score | Description |
|---|
| 0.9–1.0 | Format is native: correct length, structure, hook for the platform |
| 0.7–0.8 | Minor violations (e.g. LinkedIn post with excessive hashtags) |
| 0.5–0.6 | Functional but feels cross-posted without adaptation |
| 0.0–0.4 | Wrong format entirely (e.g. HN-formatted as a sales pitch) |
Platform norms reference:
- HN: factual title; no adjectives; technical first comment required before posting
- LinkedIn: first line works as standalone; 150–300 words for thought leadership; no engagement bait
- X/Twitter: thread tweet-1 readable alone; no emoji-only openers
- Email: subject line is the primary asset; no "Sent with [tool]" footers for developer audiences
Platform format discipline
When the destination platform is known, apply the following additional checks on top of Dimension 4 scoring. These checks are mandatory — a platform format failure drops Dimension 4 by a minimum of 0.2 per failing check.
X/Twitter
| Check | Pass | Fail |
|---|
| Hook lands in first 8 words | First 8 words carry a complete tension, fact, or claim — thread does not depend on context to make sense | Opener requires setup before the point lands; reader must read further before deciding to engage |
| Opener names a specific pain or fact | Names a specific situation, number, outcome, or mechanism | Vague generalization ("In a world where...", "A lot of people think...") |
| Thread is not pure self-promotion | Thread contains a useful angle — how-to, observation, mechanism, lesson — that has value independent of the product | Thread is entirely about the product with no useful standalone content |
| No filler affirmations | No "Here's why this matters" or "Thread 🧵" as a standalone opener | Thread starts with a meta-statement about the thread rather than the content |
IF X thread opener fails hook check AND fails specific-pain check → BLOCK. Return:
"X/Twitter opener fails two platform format checks: hook does not land in first 8 words AND opener is vague.
Both failures together significantly reduce thread performance.
Required fix: rewrite first tweet to open with [specific pain or fact] in ≤8 words."
LinkedIn
| Check | Pass | Fail |
|---|
| Professional opener (not clickbait) | Opens with a practitioner observation, question, or data point | Opens with manufactured curiosity ("You won't believe...", "I was shocked when...") |
| Opinion backed before shared | Opinion or recommendation is preceded by evidence or personal experience | Opinion stated as assertion without grounding |
| CTA is specific and low-friction | CTA names one specific action with a clear reason to take it | CTA is vague ("check it out", "let me know what you think") or asks for too much commitment from a cold reader |
| No more than one CTA per post | One next step | Multiple competing asks in a single post |
WARN if LinkedIn CTA is absent entirely. Professional practitioner posts can omit an explicit CTA, but if a CTA is present it must be specific.
Email
| Check | Pass | Fail |
|---|
| Subject line is specific | Subject line promises a concrete thing (data point, how-to, announcement with specifics) | Subject line is vague, teaser-only, or question-baiting without hinting at the answer |
| Preview text extends subject | Preview text adds new information — time context, a number, the key tension | Preview text repeats the subject line or begins with "View in browser" type boilerplate |
| First sentence earns the open | First sentence delivers on the subject line's implied promise | First sentence is setup/context before the point ("As you know, the market has been...") |
BLOCK if subject line is vague AND first sentence is setup. Both failures together = reader bails before the message lands.
Newsletter
| Check | Pass | Fail |
|---|
| Lead with insight or data | Opening paragraph contains a non-obvious insight, counter-intuitive finding, or specific data point | Opening paragraph is context-setting ("Let's talk about [category]...") |
| No 3-paragraph warm-up | Point lands within first 100 words | Three or more paragraphs of setup before the core insight appears |
| Each section has a standalone takeaway | Reader can extract a specific, usable point from each section | Section describes the topic without delivering a take |
Dimension 5 — Accuracy and compliance (0.0–1.0)
Are all claims defensible? No ToS violations, legal exposure, or unverified statistics? All proof attributed to named, verifiable sources?
| Score | Description |
|---|
| 0.9–1.0 | All claims verified; no legal exposure; metrics have methodology; proof attributed to named sources |
| 0.7–0.8 | One minor claim unverified; no compliance risk |
| 0.5–0.6 | Metrics without methodology; one comparative claim unverified |
| 0.0–0.4 | Banned hype pattern; unverified comparative claim; legal exposure; anonymous proof in customer story |
Proof attribution rule (blocking for customer stories and case studies):
- Required: named company + named role + specific metric = signal
- Block: "a large enterprise customer," "teams have seen," "our users report" without specifics
- Exception: NDA context — use role + measurable context ("the platform lead at a logistics company processing 400M events/day")
Dimension 6 — Brand voice (0.0–1.0)
Does the writing match the practitioner voice standard? (See pmm/DOMAIN.md.)
For AI-assisted drafts, authenticity is scored upstream by pmm/ai-slop-audit/SKILL.md — run it first. A draft that reads as AI (or contradicts the author's VOICE-PRINT.md) cannot pass this dimension; ai-slop-audit's verdict feeds directly into this score.
| Score | Description |
|---|
| 0.9–1.0 | Senior practitioner sharing a discovery; specific; honest about failure/limitation |
| 0.7–0.8 | Mostly practitioner; occasional slip into marketing adjective |
| 0.5–0.6 | Mixed — practitioner in some sections, marketing in others |
| 0.3–0.4 | Mostly marketing voice |
| 0.0–0.2 | Pure marketing — adjectives, hype, hollow enthusiasm |
Dimension 7 — Differentiation (0.0–1.0)
Does this content say something specific to this product, or is it generic category content?
| Score | Description |
|---|
| 0.9–1.0 | Unmistakably about this product; competitor could not use the same content |
| 0.7–0.8 | Strong product specificity; minor sections could be generic |
| 0.5–0.6 | Category content with product name inserted |
| 0.0–0.4 | Generic; any product in the category could publish this |
Dimension 8 — Actionability (0.0–1.0)
Does the reader know what to do or learn next after consuming this?
| Score | Description |
|---|
| 0.9–1.0 | Reader leaves with a concrete next step or sharpened worldview |
| 0.7–0.8 | Clear takeaway; next step slightly implicit |
| 0.5–0.6 | Informative but passive; no action implied |
| 0.0–0.4 | No takeaway; ends without pointing anywhere |
Part B — Experiment spec scoring
Use for: growth experiment specs from /growth-experiment.
Dimensions (score each 0.0–1.0)
| Dimension | Question | 0.9+ | 0.5–0.8 | 0.0–0.4 |
|---|
| Hypothesis clarity | Is the hypothesis falsifiable? Is success/failure observable? | Single sentence; specific action; named metric + threshold | Hypothesis present but metric is vague | "Improve engagement" type — not falsifiable |
| Channel fit | Do the channel's strengths match the tactic? | Channel matches growth motion; native format | Minor mismatch; likely still works | Wrong channel for motion or ICP |
| Metric alignment | Does the success metric prove or disprove the hypothesis? | Direct measurement of the hypothesis | Related metric but not the most direct | Metric does not connect to hypothesis |
| Effort vs priority | Is the effort estimate credible? Is this worth running now? | Realistic estimate; high priority constraint | Minor underestimate | Significant underestimate or low priority |
| Conversion flow | Is the path from touch → intent clear? Is friction noted? | Full path specified; known friction acknowledged | Path clear; friction not noted | No path specified |
| Attribution | Are UTMs/events defined where needed? | UTMs and events specified | Mentioned but not defined | No attribution plan |
| Resource feasibility | Are team, budget, tools, and approvals realistic? | All dependencies identified; owners assigned | Most dependencies clear | Key dependency missing or owner unassigned |
| Compliance safety | No spam, deception, or account-risk patterns? | Clean | Minor concern noted | Violation detected |
Calibrated examples
These anchors should be used to maintain score consistency across sessions.
Example 1 — APPROVE content (target avg ≥ 0.75)
Content: LinkedIn post (B2B technical product)
"We migrated [X] GB of order history from our legacy database to [Product] in a single weekend. No downtime. Here is the three-step dual-write strategy we used, and the one gotcha that nearly broke us at 2 AM."
| Dimension | Score | Notes |
|---|
| ICP alignment | 0.9 | Engineers who own database migrations; "2 AM gotcha" is visceral and specific |
| Funnel stage fit | 0.8 | Awareness/consideration; CTA is implicit (read the story) which is correct for LI |
| CTA clarity | 0.7 | No explicit CTA but the hook promises a payoff — acceptable for awareness content |
| Platform format | 0.9 | Native LinkedIn format; first line is a standalone hook |
| Accuracy & compliance | 0.9 | Specific numbers and timeframe are verifiable |
| Brand voice | 0.9 | Practitioner; specific; honest ("nearly broke us at 2 AM") |
| Differentiation | 0.7 | Dual-write is a known pattern; the "gotcha" is the differentiating hook |
| Actionability | 0.8 | Reader will read the post to get the strategy |
Average: 0.83 → APPROVE. Note: push the gotcha detail into the hook if there's room.
Calibration note: substitute your product name and real migration metric. The score pattern holds for any honest, specific, practitioner-voiced technical migration story.
Example 2 — REVISE content (target avg 0.4–0.6)
Content: Blog post intro (any technical product)
"[Product] is a game-changing [category] that seamlessly scales your workload. In this post we explore its powerful features and why it might be the best solution for your needs."
| Dimension | Score | Notes |
|---|
| ICP alignment | 0.4 | No named pain, no persona signal, no trigger context |
| Funnel stage fit | 0.6 | Blog format is right; content depth is shallow |
| CTA clarity | 0.5 | "Explore" is vague; no clear next step |
| Platform format | 0.7 | Blog format is appropriate |
| Accuracy & compliance | 0.4 | "game-changing", "seamlessly", "best solution" — three banned hype patterns |
| Brand voice | 0.3 | Pure marketing voice; no practitioner signal |
| Differentiation | 0.3 | Generic category content; any competitor could publish the same paragraph |
| Actionability | 0.4 | Reader doesn't know what to do after reading |
Average: 0.45 → REVISE. Top 3 fixes:
- Remove "game-changing", "seamlessly", "best solution" — replace with specific, verifiable claims
- Rewrite the first sentence to name a specific constraint and one concrete outcome
- Add an ICP signal: who is this for, with what pain, in what context
Example 3 — APPROVE experiment spec (target avg ≥ 0.75)
Spec: HN Show HN post for open-source infrastructure tool
"Hypothesis: If we post a Show HN on a weekday 9–11 AM EST leading with a specific migration pattern (dual-write), we will reach ≥200 upvotes and ≥30 GitHub stars within 48 hours, vs baseline of 80 upvotes for our last HN post."
"Success metric: ≥200 upvotes within 48h AND ≥30 GitHub stars within 48h"
"Distribution handoff: UTM on repo link (utm_source=hn); technical first comment drafted before posting."
| Dimension | Score | Notes |
|---|
| Hypothesis clarity | 0.9 | Falsifiable; metric numbered; baseline stated; time window clear |
| Channel fit | 0.9 | HN is the right surface for OSS infra tools |
| Metric alignment | 0.9 | Upvotes + stars directly measure reach and developer interest |
| Effort vs priority | 0.8 | Post itself is low effort; follow-up engagement in comments takes 2–3h |
| Conversion flow | 0.8 | HN post → repo → stars is clear |
| Attribution | 0.7 | UTM mentioned but not fully specified (which event tracks star click?) |
| Resource feasibility | 0.8 | Realistic; first comment needs drafting before posting |
| Compliance safety | 1.0 | Clean |
Average: 0.85 → APPROVE. Fix before posting: specify the GitHub star tracking event; draft first comment.
Calibration note: substitute your product name, real baseline, and actual target thresholds. The score pattern applies to any well-specified OSS/developer-tool distribution experiment.
Verdict logic
Average score ≥ 0.75 AND no banned hype patterns → APPROVE
"Ready to publish/run. [Optional: one improvement to consider]"
Average score 0.55–0.74 OR banned hype patterns present → REVISE
Provide top 3 specific fixes.
Re-review after fixes if score was borderline.
Average score < 0.55 OR fundamental structural problem → REJECT
"Wrong channel, non-compliant, or metric/hypothesis disconnected from the
experiment goal. Recommend rebuilding from scratch rather than patching."
Explain the structural problem and the right approach.
Output format
## Content Quality Review
**Type:** [Content piece / Experiment spec]
**Title/Name:** [Title or experiment name]
**Destination:** [Channel or platform]
**Reviewer:** content-quality-review
### Banned patterns check
[None found / List each pattern found with location]
### Dimension scores
[Part A or Part B dimensions]
Dimension 1: [score] — [one line note]
Dimension 2: [score] — [one line note]
...
Average: [calculated average]
### Verdict: [APPROVE / REVISE / REJECT]
### Fixes required (if REVISE or REJECT)
1. [Most important fix — specific and actionable]
2. [Second fix — specific and actionable]
3. [Third fix — specific and actionable]
### Notes
[Anything that scored well; patterns to repeat]
Validation criteria
Brain check (if connected)
If a companion aether-growth-brain repo is connected:
Before scoring:
- Read
playbooks/messaging.md — check if approved positioning messaging exists; score Brand voice and Differentiation dimensions against the approved messaging, not just against the content in isolation. Content that contradicts approved messaging should fail Differentiation even if it's internally coherent.
- Read
knowledge/icp-map.md — verify the content's claimed ICP matches the validated ICP; content targeting the wrong audience fails ICP alignment at 0.3 or below.
Brain not connected: proceed with content scoring; note that ICP alignment and Differentiation scores are based on the content's stated audience and claims.
Anti-patterns (reviewer anti-patterns)
| Anti-pattern | Why it fails | Fix |
|---|
| Giving 0.7 to a banned hype pattern without a rewrite note | The pattern gets through even though it fails the banned list | Banned pattern → mandatory rewrite note regardless of overall score |
| Averaging a 0.3 into an APPROVE verdict | One very bad dimension can drag the average down without triggering specific attention | Flag any dimension scoring < 0.5 as a priority fix even if the average passes |
| Vague fix notes ("make it more specific") | Author doesn't know what to change | Every fix must state: which sentence/element to change, and an example of what the replacement should look like |
| Applying B2C content standards to B2B technical audience | Emoji hooks, casual voice, and visual-first formats don't convert developers | Check ICP and platform together; developer content uses plain text and technical hooks |
| Treating Part A and Part B with identical criteria | An experiment spec is not a blog post; metric alignment matters for specs, not for content | Apply Part A criteria to content pieces and Part B criteria to experiment specs only |
| Scoring without calibrated examples as anchors | Scores drift across sessions; APPROVE/REVISE threshold becomes inconsistent | Always reference the three calibrated examples before finalizing scores |
| Platform-agnostic review misses platform-specific quality standards | Dimension 4 scores "format is appropriate" without checking platform-specific hook rules, opener quality, CTA type, or structural requirements — a LinkedIn post with a vague CTA and a manufacturing hook passes generic review but fails in the field | When destination platform is known, run Platform format discipline checks before finalizing Dimension 4 score |
Benchmarks (content performance, 2025–2026)
| Channel | Benchmark | Notes |
|---|
| Email subject line open rate (developer tools, cold list) | 15–25% | Warm list: 30–45% |
| Email CTA click rate (developer tools) | 3–8% | Plain text outperforms HTML by 15–20% |
| LinkedIn thought leadership post engagement rate | 1–5% | First 60 minutes engagement determines reach |
| HN Show HN — qualifying upvote threshold | >20 upvotes to front page | Requires strong technical first comment |
| Blog post → trial signup conversion (SEO traffic) | 1–3% | Technical depth is the primary driver |
| Landing page headline test: specific vs generic | Specific outperforms generic by 20–40% | Field benchmark (developer tools context) |
| Content with banned hype pattern conversion penalty | -15 to -30% vs equivalent non-hype copy | Trust signal research, Nielsen 2025 |
Related skills
| Skill | When to use |
|---|
pmm/positioning-review/SKILL.md | Before reviewing content: positioning must be FINAL for Layer 3 copy |
pmm/launch/SKILL.md | After APPROVE: launch workflow requires content-quality-review APPROVE before Phase 2 |
pmm/launch-readiness-review/SKILL.md | S2 check: this review's APPROVE verdict enables the launch readiness GO |
growth/growth-experiment/SKILL.md | Part B scoring: experiment spec review |
pmm/DOMAIN.md | Practitioner voice standard, AI-specific banned phrases, five story arcs |
References & Sources
Tier 2 (operator templates — adapted, not authoritative):
- b2b-practitioner-voice (growth-skills v1.0, score 9/10): practitioner voice standard, five story arcs, context-before-code rule
- growth-quality-rubric (growth-skills v1.0, score 9/10): 0.0–1.0 scoring rubric, APPROVE/REVISE/REJECT framework, calibrated example structure
- growth-publish-safety (growth-skills v1.0, score 6.5/10): banned hype pattern list
