# SEO & GEO Standards for TiDB Content

**Owner:** PMM  
**Update trigger:** Search algorithm shifts; new AI citation research; quarterly content audit  
**Applies to:** All externally published TiDB content — blog posts, case studies, technical articles, thought leadership pieces

---

## What this file covers

Universal SEO and GEO (Generative Engine Optimization) principles that apply across all TiDB content types. Skill-specific patterns (e.g., case study schema markup, developer community norms) live inside each skill. Load this file alongside any content-creation skill for distribution-ready output.

---

## Headline formula

**General formula:** `[Subject or problem] + [Specific outcome or insight] + [Context or qualifier (optional)]`

Pass/fail test — ask: "Would an engineer bookmark this title in search results?"

| Format | Weak | Strong |
|--------|------|--------|
| Migration story | "How [Company] Uses TiDB" | "How [Company] Eliminated 2 AM Pages by Moving Off MySQL Sharding" |
| Technical how-to | "Using TiDB for AI Workloads" | "Building Real-Time AI Agents on TiDB: One Database, No ETL Pipeline" |
| Architectural decision | "TiDB vs Aurora" | "Why We Moved 500M Rows from Aurora to TiDB (and What We'd Do Differently)" |
| Thought leadership | "AI and Databases" | "The Consistency Trap in AI Agent State — and Why Most Backends Will Fail at Scale" |
| Case study | "TiDB Case Study: [Company]" | "[Company]: 30% Lower Operational Cost with Zero Downtime Using TiDB" |

**Rule:** Lead with the subject or outcome, not with "TiDB" or "PingCAP". The reader should recognize their problem before they see the product name.

---

## GEO: Optimizing for AI citation engines

AI engines (Perplexity, ChatGPT with search, Google AI Overviews, Claude) select passages to cite based on **information density, specificity, and definitional clarity**. To maximize citation probability:

### 1. Definition-first opening

The first sentence of any article should be a self-contained, definitional statement about the subject and the challenge. AI engines are trained to surface this pattern as a direct answer.

**Pattern:** `[Subject] is a [what it is] [doing/serving scale indicator]. When [specific challenge], [what happened or was done].`

**Pass:** "BookMyShow is India's largest online entertainment ticketing platform, processing millions of transactions during peak events. When their database couldn't handle traffic spikes without manual intervention, they migrated to TiDB — cutting operational costs by 30%."

**Fail:** "In today's fast-paced digital environment, companies face many database challenges."

### 2. Statistic density

Target **2-3 quantified data points per 300 words**. AI engines weight specific, verifiable claims over vague assertions.

- "Tables exceeding 2 billion records caused query timeouts during peak hours" → citable
- "They had scaling issues" → not citable

Format: `[metric] + [workload or context] + [before/after or consequence]`

### 3. Entity clarity

Mention proper names (TiDB, customer company, technology stack) by their **full, unambiguous proper name on first use**. AI systems resolve entities more reliably with clear references.

- First mention: "TiDB Cloud Dedicated" not "TiDB"
- First mention: "Amazon Aurora (MySQL-compatible)" not "Aurora"
- Use `@brain/knowledge/product.md` for official TiDB product and tier names

### 4. Quote inclusion

Direct quotes from named sources improve citation rates by adding a credibility signal AI models weight. Even one well-placed quote significantly increases the chances of passage citation.

Format: pull quote immediately after the result section it supports, with full name and title attribution.

### 5. Freshness

Citation rates drop sharply after content goes stale. Two rules:
- Include a visible **"Published"** and **"Last updated"** date on all published content
- **Review and update within 90 days** of publish; set a calendar reminder. Pages updated within 2 months earn significantly more AI citations than older content

### 6. Answer extraction (per-section)

The definition-first opening (rule 1) optimizes the article's top. This rule optimizes the *body*: AI engines extract answers section by section, not just from the intro.

- **Self-contained answer in the first 2-3 sentences under every H2.** An engine should be able to lift those opening sentences and have a complete answer to the question the H2 implies. Put the direct answer first, then expand — don't bury it under setup.
- **FAQ section for "what is / how do I / comparison" posts.** Three to six Q&As, each answer two to four sentences. Pull questions from Google's "People Also Ask" for the target keyword, or ask "what would a reader still wonder after finishing this?" Mark up with **FAQ schema (JSON-LD)** so the answers are eligible for rich results and AI extraction.
- **Table of contents with jump links** for posts over ~1,500 words. Helps both human skimmers and crawlers map the structure.

---

## Traditional SEO: on-page essentials

### Keyword placement

- **Primary keyword** in: title, first paragraph, one H2 header, meta description
- **Secondary keywords** woven naturally into body text — don't force them
- Target **long-tail keyword clusters**, not head terms:
  - `[industry] + distributed SQL / MySQL alternative / database scaling`
  - `TiDB + [use case]` (e.g., "TiDB for fintech", "TiDB HTAP")
  - `[competitor] migration or alternative` (e.g., "migrate from Aurora to TiDB")

### Meta description

- Under 160 characters
- Lead with the specific outcome or insight, not with "In this article..."
- Include: primary keyword + one concrete result or claim

**Pattern:** `[Customer or subject] [achieved/solved] [specific outcome] with TiDB. [One-line context].`

### Internal linking

Every published piece should link to:
- The TiDB product page for the specific product mentioned
- 1-2 related pieces covering the same industry or use case (topical authority clustering)
- One upstream resource (documentation, benchmark, or deep-dive) for technical readers who want to go deeper

**Minimum three internal links** per post (five to eight for longer pieces). Beyond the targets above:
- **Descriptive anchor text** — the anchor names the destination ("the TiDB vs. Aurora comparison"), never "click here" or "this post". Both readers and crawlers use anchor text to understand the linked page.
- **Named-entity link pass.** On a final pass, check every named entity — product, feature, customer, competitor, concept. If a relevant internal page exists (product page, docs page, case study, comparison), link it. If none exists, flag it as a content gap rather than forcing a link.
- **No link farming.** If a link doesn't help the reader, remove it.

### URL structure

- Use hyphens, not underscores
- Include the primary keyword in the slug
- Keep slugs under 60 characters where possible

---

## Content type quick-reference

| Content type | Primary SEO goal | Key GEO element |
|-------------|-----------------|-----------------|
| Case study | Rank for `[industry] + TiDB` and `[competitor] migration` | Definition-first opening + stat density in results section |
| Technical article / tutorial | Rank for `[use case] + TiDB` and `how to [task]` | Code examples with explicit context; specific version numbers |
| Thought leadership | Earn backlinks and brand authority; secondary keyword target | Named claim in first paragraph; direct quote from named source |
| Competitive content | Rank for `TiDB vs [competitor]` | Fair comparison with specific constraint evidence |