---
name: tidb-content-review
description: >-
  Final quality gate for any TiDB external content before publishing. Scores
  content across 9 dimensions (claim accuracy, messaging alignment, banned
  phrases, brand voice, channel fit, ICP relevance, honesty, CTA clarity, CTA
  appropriateness) and outputs Approve / Revise / Reject with specific fixes.
  Use before publishing any blog post, social copy, benchmark result, case
  study, event content, or sales material. No content ships without clearing
  this gate.
---

# TiDB content review

**Role:** Publication Gate. You are the last check before content reaches the public. Your job is not to make content better (that happened upstream) — it is to confirm that the content meets the bar to represent TiDB externally. One poorly framed benchmark or one overclaimed compatibility statement creates more damage than not publishing at all.

---

## Who runs this review

| Content type | Minimum reviewer |
|-------------|-----------------|
| Technical blog / case study | PMM + (PM or DevRel) |
| Benchmark content | PMM + PM + (Legal if competitive claims) |
| Thought leadership | PMM |
| Social copy | PMM or designated content lead |
| Event / field content | PMM + Field Marketing lead |
| Developer/community content | DevRel lead (community content) or PMM (owned channels) |
| Sales enablement content | PMM + Sales lead |

---

## 8 review dimensions

Score each 0.0–1.0. A critical issue in any dimension is a block regardless of score. An average below 0.6 on any single dimension = Revise.

---

### Dimension 1: Claim accuracy

**Question:** Are all technical claims grounded in verified evidence?

| Score | Criteria |
|-------|---------|
| 0.9–1.0 | All claims trace to a named proof point in `tidb-messaging-platform`; methodology disclosed for benchmarks |
| 0.7–0.8 | Most claims grounded; 1–2 claims need citation added but no overclaims |
| 0.5–0.6 | Some claims are directionally true but overstated; benchmark without full methodology |
| 0.0–0.4 | **Critical:** Unverifiable claims; fabricated data; benchmark contamination; customer reference without permission |

**Critical checks:**
- [ ] No claim of "MySQL compatible" without the protocol compatibility caveat and link to docs
- [ ] No performance numbers without workload context (TPS, rows, nodes, version)
- [ ] No customer case studies without documented customer approval
- [ ] No roadmap features presented as GA
- [ ] TiDB version specified for version-dependent claims
- [ ] **Customer stories / case studies:** all proof attributed to named companies, named roles, and specific metrics — blocking if "a large enterprise customer" or "a team of 50 engineers" without explicit anonymization justification

---

### Dimension 2: Messaging alignment

**Question:** Does the content align with the approved messaging platform?

| Score | Criteria |
|-------|---------|
| 0.9–1.0 | All TiDB-specific claims trace to `tidb-messaging-platform` approved proof points; positioning matches approved statement |
| 0.7–0.8 | Core positioning aligned; minor claims not in platform but not contradicting it |
| 0.5–0.6 | Some messaging is off-brand or uses outdated positioning (e.g., HTAP-centric framing when AI-native is current) |
| 0.0–0.4 | **Critical:** Claims that contradict the approved messaging platform; introduces unapproved competitive comparisons |

---

### Dimension 3: Banned phrases check

**Question:** Is the content free of all Tier 1 banned phrases?

Tier 1 (block — must remove before publishing):
- [ ] "fully MySQL compatible" (not: "MySQL protocol compatible")
- [ ] "zero downtime" without specification
- [ ] "seamless", "frictionless", "effortless"
- [ ] "game-changing", "revolutionary", "industry-leading", "best-in-class"
- [ ] "AI-ready" without specific capability
- [ ] "scales infinitely"
- [ ] Any benchmark claim without methodology
- [ ] Customer name used without documented approval

Tier 2 (advisory — improve before major campaign):
- [ ] "powerful" without proof → replace with specific capability
- [ ] "high performance" without numbers
- [ ] "leverage" → use "use"
- [ ] "HTAP" without explanation for non-technical audience

---

### Dimension 4: Brand voice consistency

**Question:** Does the content sound like TiDB, as defined in `tidb-brand-voice`?

| Score | Criteria |
|-------|---------|
| 0.9–1.0 | Peer-engineer voice; specific; honest about limitations; no superlatives; practitioner credibility |
| 0.7–0.8 | Mostly correct tone; one or two marketing phrases that weaken credibility |
| 0.5–0.6 | Reads like marketing copy; adjective-heavy; trust signals missing |
| 0.0–0.4 | **Critical:** Corporate press release voice; hollow hype; anti-patterns throughout |

---

### Dimension 5: Channel fit

**Question:** Is the format appropriate for the distribution channel?

| Score | Criteria |
|-------|---------|
| 0.9–1.0 | Length, structure, and format match the platform (e.g., HN post has methodology in first comment; LinkedIn has hook + line breaks; WeChat has image and sections) |
| 0.7–0.8 | Generally appropriate; minor format issues |
| 0.5–0.6 | Content doesn't match platform norms (e.g., 500-word block paragraph for Twitter; press-release language on HN) |
| 0.0–0.4 | **Critical:** Wrong format entirely; HN post with no methodology; WeChat with no image |

---

### Dimension 6: ICP relevance

**Question:** Does the content speak to the target ICP's actual pain and use deciding language?

| Score | Criteria |
|-------|---------|
| 0.9–1.0 | Opens with ICP pain; uses deciding language from ICP card; specific use case for the target segment |
| 0.7–0.8 | Relevant to the ICP but generic; missing ICP-specific language |
| 0.5–0.6 | Could apply to any database audience; no segment specificity |
| 0.0–0.4 | **Critical:** Targeting the wrong audience for the channel or content type; e.g., enterprise migration story on AI developer forum |

**Advisory — Problem naming:**
Does the content coin or use a specific, memorable name for the problem it solves? Named problems ("staging drift," "MySQL write ceiling," "HTAP tax") are more ICP-resonant and memorable than generic descriptions. Not blocking, but flag as an improvement opportunity when absent from technical or tutorial content.

---

### Dimension 7: Honesty (the trust dimension)

**Question:** Does the content acknowledge limitations, trade-offs, or challenges?

| Score | Criteria |
|-------|---------|
| 0.9–1.0 | Explicitly acknowledges limitations; shows a challenge or failure; states what TiDB is NOT the right choice for |
| 0.7–0.8 | Mostly honest; one limitation could be more explicit |
| 0.5–0.6 | All positive; no limitations mentioned; feels cherry-picked |
| 0.0–0.4 | **Critical:** Benchmark suppresses failure cases; customer story omits significant caveats; technical claim is provably incomplete |

*Note: For engineering audiences, stating limitations increases credibility. Suppressing them decreases it and invites community correction.*

**Additional critical checks for this dimension:**
- [ ] **Tutorial / comparison content:** alternatives presented honestly — each with legitimate strengths before limitations disclosed. Strawmanned alternatives are a trust-destroying pattern with engineer audiences; treat as a critical issue blocking publication.
- [ ] **Technical content:** every TiDB feature mentioned connects to a specific workflow example — not a naked feature list. Advisory for thought leadership; blocking for tech stories and developer tutorials.

---

### Dimension 8: CTA clarity

**Question:** Is the reader's next step clear, appropriate, and achievable?

| Score | Criteria |
|-------|---------|
| 0.9–1.0 | One CTA; specific and achievable; appropriate for the content type and funnel stage |
| 0.7–0.8 | CTA present but generic ("learn more") |
| 0.5–0.6 | No clear CTA; or CTA is mismatched to funnel stage (e.g., "request a demo" for awareness content) |
| 0.0–0.4 | Multiple conflicting CTAs; or CTA undermines the piece (e.g., hard sell at end of thought leadership essay) |

---

### Dimension 9: CTA appropriateness

**Question:** Is the CTA *type* appropriate for the content type, audience, and trust context?

This dimension is separate from Dimension 8. Clarity (D8) asks "is the CTA clear?" — appropriateness (D9) asks "should there be an explicit CTA at all, and is this the right kind for this audience?"

| Score | Criteria |
|-------|---------|
| 0.9–1.0 | CTA type matches content type and audience: developer/engineering trust content uses implicit CTA or progressive low-friction CTA; campaign content uses explicit CTA |
| 0.7–0.8 | CTA is present and functional but slightly mismatched to content type |
| 0.5–0.6 | CTA mismatched: explicit hard sell on developer content; or no CTA on a campaign landing page |
| 0.0–0.4 | **Critical:** Urgent conversion language ("act now," "limited time") or hard demo-request CTA on trust-building content for engineering audiences — actively damages credibility with this audience |

**CTA calibration by content type:**

| Content type | Appropriate CTA approach |
|-------------|------------------------|
| Technical blog / migration story | Implicit (scale proof as closer) or low-friction ("explore the docs," "try for free") |
| Thought leadership essay | Implicit — a vision statement or call to reflection; NOT "request a demo" |
| Customer case study | Implicit (proof-of-adoption metric as closer); or soft "read more stories" |
| Partner spotlight | Implicit or partner-specific ("try [partner] powered by TiDB") |
| Channel copy for campaigns | Explicit CTA required; progressive commitment preferred (free tier → upgrade) |
| Developer community / HN / GitHub | No CTA or soft invite to engage; hard sell is reputation damage |

**Progressive CTA design (advisory):** When an explicit CTA is appropriate, design with progressive commitment: low friction (try for free, read docs) → medium (join community) → high (request demo). Don't lead with the highest commitment ask.

---

## Review output format

```markdown
## Content Review — [Title] — [Date] — [Channel]

### Aggregate score
| Dimension | Score | Issues |
|-----------|-------|--------|
| Claim accuracy | [0.0-1.0] | [None / specific issue] |
| Messaging alignment | [0.0-1.0] | [None / specific issue] |
| Banned phrases | [PASS / FAIL] | [None / list blocked phrases found] |
| Brand voice | [0.0-1.0] | [None / specific issue] |
| Channel fit | [0.0-1.0] | [None / specific issue] |
| ICP relevance | [0.0-1.0] | [None / specific issue] |
| Honesty | [0.0-1.0] | [None / specific issue] |
| CTA clarity | [0.0-1.0] | [None / specific issue] |
| CTA appropriateness | [0.0-1.0] | [None / specific issue] |

### Verdict: APPROVE / REVISE / REJECT

### Blockers (must fix before publishing)
- [Dimension]: [specific text] → Required change: [what to write instead]

### Advisory (fix before major campaign, optional for single posts)
- [Dimension]: [specific text] → Suggested improvement: [what to change]

### Reviewer: [name] — [Date]
```

---

## Calibration examples

### Example: APPROVE (avg 0.84)

> LinkedIn post: "We hit MySQL write saturation at 8k TPS on Black Friday. Here's the dual-write migration strategy we used to move 800GB to TiDB with no downtime window, and the one schema assumption that nearly broke us at 2 AM."
>
> - Claim accuracy 0.9 — "no downtime" is a strong claim; acceptable if the body explains the maintenance-window-free approach with a dual-write strategy
> - Messaging alignment 0.9 — MySQL scale-out migration story matches approved proof points
> - Banned phrases PASS — "no downtime" is borderline but acceptable with technical explanation
> - Brand voice 0.9 — practitioner voice, specific metric, honest struggle
> - Channel fit 0.9 — strong hook, appropriate for LinkedIn
> - ICP relevance 0.9 — directly addresses MySQL scale-out ICP
> - Honesty 0.8 — shows the "nearly broke us" moment; could be more explicit about what TiDB gave up
> - CTA clarity 0.7 — implied CTA via link; could be more explicit
>
> **APPROVE.** Advisory: add explicit CTA. Consider adding one-sentence limitation disclosure.

### Example: REVISE (avg 0.45)

> Blog post intro: "TiDB is an AI-native database that seamlessly handles any AI workload with industry-leading performance and effortless scalability. In this post, we explore TiDB's powerful features for modern AI applications."
>
> - Claim accuracy 0.3 — "any AI workload" and "industry-leading" are unverifiable
> - Messaging alignment 0.4 — uses outdated generic framing, not AI-agent specific
> - Banned phrases FAIL — "seamlessly", "industry-leading", "powerful", "effortless" all blocked
> - Brand voice 0.2 — marketing copy, not practitioner voice
> - Channel fit 0.6 — blog format is fine; content is the problem
> - ICP relevance 0.3 — no specific ICP pain or segment identified
> - Honesty 0.2 — no limitations, no trade-offs, no specificity
> - CTA clarity 0.5 — "explore features" is too vague
>
> **REVISE.** Rewrite the intro to name a specific AI workload pain. Remove all Tier 1 banned phrases. Add a "who this is for" sentence. Add one limitation disclosure.

---

## Pairs with

- `tidb-brand-voice` — banned phrases and voice standards
- `tidb-messaging-platform` — proof point table and approved claims
- `tidb-benchmark-narrative` — benchmark methodology standards
- `tidb-tech-story` — writing mechanics for technical content
- `tidb-thought-leadership` — thought leadership rubric
- `tidb-channel-copy` — channel format standards
- `@tidb-gtm-brain/playbooks/messaging.md` — approved messaging quick reference