product-help-center（product help center 搜索）

Help Center Design

Design AI-first help centers, knowledge bases, FAQs, and learning materials.

This skill reflects the shift from static help portals to AI-powered, embedded, personalized self-service systems.

Workflow (Use As Default Order)

1. Define scope and constraints

- Audience/personas, product area(s), product versioning, channels (web/in-app), compliance requirements, localization needs.

1. Inventory current knowledge

- Top tickets, top searches, top articles, top escalation reasons, and known content owners.

1. Build information architecture

- Category structure, tagging, navigation, URL strategy, and internal linking.

1. Standardize content

- Article types, templates, AI-friendly writing rules, and visual standards.

1. Instrument and measure

- KPIs, event tracking, dashboards, and search query logging.

1. Add AI support safely

- Retrieval-first answers, citations, confidence thresholds, escalation rules, and transactional guardrails.

1. Run knowledge operations

- Governance, freshness detection, release-driven updates, and continuous optimization.

Expected outputs (adapt to request):

• Help center taxonomy map + tag schema
• Top 20 article backlog (by impact) + templates
• Analytics spec (events + dashboard KPIs)
• AI support spec (RAG sources, escalation thresholds, safety rules)
• Operating cadence (owners + review schedule)

Quick Reference

Content Type Decision Matrix

Platform Selection (Verify Pricing And Plan Limits)

See references/platform-guides.md for setup/migration notes and data/sources.json for curated comparison sources.

2025-2026 Best Practices

Key Shifts

Avoid quoting hard statistics without verification; refresh trends and benchmarks via data/sources.json when needed.

AI-First Principles

1. Agentic Resolution — AI executes tasks (refunds, bookings, updates), not just answers
2. Semantic Understanding — Intent-based search, not keyword matching
3. Proactive Assistance — Surface help before users ask
4. Content Freshness — Auto-detect stale content, suggest updates
5. Multi-Source Synthesis — Pull from docs, tickets, Slack, release notes
6. Memory-Rich AI — Retain context across sessions for personalized support

Emerging Trends (2026)

Help Center Architecture

Category Structure Rules

HIERARCHY LIMITS
- Maximum depth: 2-3 levels
- Top-level categories: 5-9 (cognitive load principle)
- Articles per category: 10-20 (scannable)
- Avoid: Deep nesting, internal org structure

Recommended Top-Level Categories

STANDARD CATEGORIES (adapt to product)
1. Getting Started        — First-run, setup, quick wins
2. [Core Feature 1]       — Primary use case
3. [Core Feature 2]       — Secondary use case
4. Account & Billing      — Settings, payments, security
5. Integrations           — Third-party connections
6. Troubleshooting        — Common issues, error codes
7. API & Developers       — Technical documentation
8. What's New             — Changelog, releases

Navigation Patterns

• Breadcrumbs — Always show location in hierarchy
• Related Articles — 3-5 contextually relevant links
• Next Steps — Guide to logical next action
• Search Prominence — Above fold, always visible
• Popular Articles — Surface high-traffic content

Article Types (Keep The Set Small)

• How-To: task completion, 3-10 steps
• Troubleshooting: symptoms -> causes -> solutions
• FAQ: fast answers with links to deeper docs
• Conceptual: explain terms and mental models
• Reference: precise specs (tables, limits, error codes)

Use the copy-paste templates in references/article-templates.md.

AI Integration Patterns

Chatbot Architecture

MODERN AI SUPPORT FLOW (2025)

User query
  -> Intent detection (semantic understanding)
  -> RAG retrieval (KB + tickets + docs)
  -> Response and action (answer and/or execute task)
  -> Escalation check (confidence below threshold?)
  -> Human agent (if needed)

Agentic AI Capabilities (2025-2026)

Content for AI Consumption

AI-FRIENDLY WRITING RULES

DO:
- Clear headings with keywords
- Structured data (tables, lists)
- Explicit step numbering
- Error messages verbatim
- Unique article titles

DON'T:
- Ambiguous pronouns
- Implicit assumptions
- Marketing fluff in support content
- Duplicate content across articles

See references/ai-integration.md for RAG setup, evaluation, and escalation patterns.

Metrics & KPIs

Core Metrics

Content Health Metrics

FRESHNESS INDICATORS
- Last updated > 6 months -> Review required
- Last updated > 12 months -> Likely stale
- No views in 90 days -> Consider archive
- High bounce rate -> Content mismatch

QUALITY INDICATORS
- Thumbs down > 20% -> Rewrite needed
- Escalation after viewing -> Content gap
- Search -> immediate exit -> Title mismatch

ROI Calculation

SELF-SERVICE ROI FORMULA

Monthly Savings = (Deflected Tickets x $13) - Platform Cost

Example:
- 1,000 deflected tickets/month
- $13 average agent cost
- $500 platform cost
- ROI = ($13,000 - $500) = $12,500/month

See references/metrics-optimization.md for instrumentation, dashboards, and optimization playbooks.

Learning & Onboarding

In-App Help Patterns

Tutorial Best Practices (2025)

VIDEO TUTORIALS
- Length: 2-4 minutes (40% higher completion)
- Format: Screen recording + voiceover
- Chapters: Clickable sections
- Captions: Always include (accessibility)

INTERACTIVE GUIDES
- Click-through walkthroughs
- Sandbox environments
- Progress saving
- Skip option for experienced users

See references/learning-paths.md for onboarding sequence design, accessibility, and measurement.

Knowledge Operations (2026)

Operate the help center like a product:

• Assign owners per category and per top article; define review cadence and SLAs for updates.
• Use release notes, incident reports, and ticket trends as automatic triggers for content updates.
• Use freshness signals (search exits, escalation after article view, downvotes) to prioritize rewrites.

See references/knowledge-ops.md for governance, workflows, and checklists.

Implementation Checklist

Phase 1: Foundation (Week 1-2)

REQUIRED:

• Choose platform (Zendesk/Intercom/Freshdesk)
• Define category structure (5-9 top-level)
• Create article templates for each type
• Set up analytics tracking
• Configure search settings

Phase 2: Content (Week 3-4)

REQUIRED:

• Audit existing documentation
• Migrate/rewrite top 20 articles
• Add visual content (screenshots, GIFs)
• Implement internal linking
• Set up redirects from old URLs

Phase 3: AI Integration (Week 5-6)

REQUIRED:

• Enable AI chatbot
• Configure RAG/semantic search
• Set escalation thresholds
• Test common queries
• Monitor resolution rates

Phase 4: Optimization (Ongoing)

REQUIRED:

• Review zero-result searches weekly
• Update stale content monthly
• A/B test article titles
• Analyze escalation patterns
• Expand based on ticket trends

Resources

Trend Awareness Protocol

REQUIRED: When users ask recommendation questions about help centers, knowledge bases, or support platforms, run a quick web search to confirm current trends before answering. Prefer sources flagged add_as_web_search: true in data/sources.json, plus official docs for any platform you recommend.

Trigger Conditions

• "What's the best help center platform?"
• "What should I use for [knowledge base/FAQ/support]?"
• "What's the latest in customer self-service?"
• "Current best practices for [AI support/chatbots]?"
• "Is [Zendesk/Intercom/Freshdesk] still relevant in 2026?"
• "[Zendesk] vs [Intercom] vs [other]?"
• "Best AI chatbot for customer support?"

Required Searches

1. Search: "help center best practices 2026"
2. Search: "[specific platform] vs alternatives 2026"
3. Search: "AI customer support trends January 2026"
4. Search: "knowledge base platforms 2026"

What to Report

After searching, provide:

• Current landscape: What support platforms/tools are popular NOW
• Emerging trends: New AI capabilities, patterns, or platforms gaining traction
• Deprecated/declining: Approaches or tools losing relevance
• Recommendation: Based on fresh data, not just static knowledge

If web search is unavailable, state that constraint and proceed with best-effort static guidance.

Example Topics (verify with fresh search)

• Help center platforms (Zendesk, Intercom, Freshdesk)
• AI support agents (Fin AI, Ada, Forethought)
• Knowledge base tools (Document360, GitBook, Notion)
• In-app guidance (UserPilot, Pendo, Chameleon)
• Self-service AI capabilities and resolution rates
• Semantic search and RAG for support

Fact-Checking

• Use web search/web fetch to verify current external facts, versions, pricing, deadlines, regulations, or platform behavior before final answers.
• Prefer primary sources; report source links and dates for volatile information.
• If web access is unavailable, state the limitation and mark guidance as unverified.