Support AI Skill
Support Knowledge Management
Build and maintain an internal knowledge management system for support agents, ensuring consistent, accurate, and accessible information across the team. Use when creating agent knowledge bases, managing internal documentation, building runbooks and playboo...
Support Knowledge Management & Organization
Build and maintain an internal knowledge management system that gives support agents instant access to accurate, organized, and actionable information — reducing handle time, improving consistency, and enabling faster onboarding.
Workflow
- Audit existing knowledge sources (documentation, wikis, shared drives, tribal knowledge).
- Design knowledge taxonomy aligned to support workflow and ticket categories.
- Create knowledge creation and governance workflow (create, review, approve, publish, retire).
- Implement knowledge platform (internal wiki, KB tool, or integrated help desk knowledge).
- Train agents on knowledge usage and contribution.
- Measure knowledge effectiveness (usage, deflection, agent satisfaction).
- Maintain quality through review cycles, staleness detection, and feedback loops.
Knowledge Architecture
INTERNAL KNOWLEDGE TAXONOMY
=============================
Category 1 — Product Knowledge (40% of articles):
→ Feature guides: How each feature works, use cases, limitations
→ Troubleshooting guides: Common issues + solutions for each product area
→ Known issues: Current bugs, workarounds, ETA for fixes
→ Release notes: New features, changes, deprecations (current + last 3 versions)
→ Comparison: Product tier differences, feature availability by plan
→ Integrations: Setup guides, troubleshooting for each integration
Category 2 — Process & Procedures (25% of articles):
→ Standard operating procedures (SOPs): Step-by-step for common processes
→ Escalation procedures: When and how to escalate, who to contact
→ Refund/credit policies: Approval process, amounts, documentation
→ Data handling: How to access, export, delete customer data
→ SLA definitions: Response and resolution targets by tier and priority
→ Compliance procedures: GDPR, CCPA, SOC2 response procedures
Category 3 — Customer & Account (15% of articles):
→ Account types: Tiers, features, pricing, contract details
→ Common customer setups: Typical configurations by industry/size
→ VIP customer profiles: Key enterprise accounts, special handling notes
→ Billing scenarios: Common billing questions and resolutions
→ Churn risk indicators: What to watch for, how to flag
Category 4 — Tools & Systems (10% of articles):
→ Help desk guide: How to use Zendesk/Freshdesk (for new agents)
→ CRM procedures: Salesforce navigation, data entry standards
→ Internal tools: Custom scripts, dashboards, utilities
→ API documentation: Internal APIs, testing endpoints, authentication
→ Automation: Available macros, automations, triggers
Category 5 — Communication Templates (10% of articles):
→ Response templates: By scenario (apology, escalation, follow-up, etc.)
→ Email templates: Proactive outreach, status updates, announcements
→ Chat scripts: Opening, troubleshooting, closing scripts
→ Call scripts: Phone support framework
→ Escalation communications: How to communicate escalation to customer
KNOWLEDGE ARTICLE STRUCTURE:
→ Title: Searchable, action-oriented ("How to reset user permissions")
→ Summary: 2–3 sentences (visible in search results)
→ Last reviewed: Date + reviewer name
→ Author: Content owner
→ Tags: Category, product area, audience, version
→ Related articles: Cross-references
→ Attachments: Screenshots, videos, reference documents
→ Version history: Track changes
Knowledge Governance
KNOWLEDGE GOVERNANCE WORKFLOW
===============================
Roles and Responsibilities:
════════════════════════════════════════════════════════════════════════
Role | Responsibility
════════════════════════════════════════════════════════════════════════
Knowledge Manager | Overall governance, taxonomy, quality standards,
| review cadence, platform management
Content Owners | Domain experts responsible for specific categories
| (1 per major category: Product, Process, etc.)
Contributors | Support agents who create and update articles
| in their areas of expertise
Reviewers | Senior agents/leads who review articles for accuracy
| before publication
All Agents | Use knowledge base, flag outdated content,
| suggest improvements
════════════════════════════════════════════════════════════════════════
Content Lifecycle:
1. CREATE: Agent identifies knowledge gap → drafts article
→ Template ensures consistent structure
→ Tags and metadata required
2. REVIEW: Content owner reviews within 3 business days
→ Accuracy check: Information correct and complete?
→ Clarity check: Easy to understand and follow?
→ Searchability check: Will agents find it when searching?
→ Approve / Request Changes / Reject
3. PUBLISH: Approved article published to knowledge base
→ Visible to all agents immediately
→ Notification: "New article published: [title]"
4. MAINTAIN: Quarterly review cycle
→ Content owner reviews all articles in their category
→ Staleness flag: Articles not reviewed in 90 days
→ Update / Retire / Archive
5. RETIRE: Outdated or superseded articles retired
→ Redirect to replacement article
→ Archive for reference (not hidden, marked "archived")
→ Notification: "Article [title] has been updated; see new version"
KNOWLEDGE QUALITY STANDARDS:
════════════════════════════════════════════════════════════════════════
Criterion | Standard
════════════════════════════════════════════════════════════════════════
Accuracy | Fact-checked, tested, verified
Completeness | Covers the full scope of the topic
Current | Reviewed within last 90 days
Clear | 8th grade reading level
Actionable | Step-by-step instructions, not just description
Searchable | Title and tags match agent search language
Consistent | Follows knowledge base style guide
Linked | Related articles referenced
════════════════════════════════════════════════════════════════════════
Knowledge Platform and Usage
KNOWLEDGE PLATFORM SELECTION
==============================
Option 1 — Help Desk Integrated Knowledge:
→ Zendesk Guide (internal + external), Freshdesk Articles
→ Pros: Same platform as tickets, context-aware suggestions
→ Cons: Limited customization, platform-specific
→ Best for: Teams already on premium help desk plans
Option 2 — Dedicated Internal Wiki:
→ Confluence, Notion, Slab, Guru
→ Pros: Rich editing, version control, powerful search
→ Cons: Separate platform, context switching
→ Best for: Teams wanting robust documentation platform
Option 3 — Custom Knowledge Platform:
→ Build: Custom wiki with help desk integration
→ Pros: Full customization, deep integration
→ Cons: Development effort, maintenance
→ Best for: Large teams with unique needs
KNOWLEDGE PLATFORM MUST-HAVES:
→ Full-text search (with synonym support)
→ Article versioning and history
→ Content review workflows (create → review → approve → publish)
→ Staleness detection (flag old articles)
→ Usage analytics (most viewed, zero-view articles)
→ Agent feedback (helpful/not helpful ratings)
→ Mobile-friendly (agents on-the-go)
→ Browser extension or sidebar (quick access from help desk)
USAGE ANALYTICS:
════════════════════════════════════════════════════════════════════════
Metric | Target
════════════════════════════════════════════════════════════════════════
Knowledge articles viewed/week | 500+ (indicates active usage)
Articles with zero views (> 90 days)| < 10% (remove or improve)
Avg article view time | > 30 seconds (actual reading)
Knowledge-assisted tickets | > 60% (articles referenced during resolution)
Agent knowledge satisfaction | > 4.0/5.0
Knowledge gap reports/week | 5–10 (indicates active contribution)
════════════════════════════════════════════════════════════════════════
TOP 10 MOST VIEWED ARTICLES:
1. Refund and credit policy (156 views/week)
2. Escalation procedure (132 views/week)
3. Password reset steps (98 views/week)
4. Integration troubleshooting checklist (87 views/week)
5. Response templates library (76 views/week)
6. Known issues - current sprint (65 views/week)
7. GDPR data request procedure (54 views/week)
8. API rate limiting guide (48 views/week)
9. Billing cycle explanation (42 views/week)
10. Account tier comparison (38 views/week)
Integration Points
- Help Desk (Zendesk, Freshdesk, Intercom): Knowledge sidebar, article suggestions during ticket handling
- Internal Wiki (Confluence, Notion, Guru, Slab): Primary knowledge platform, version control, workflows
- Communication (Slack, Teams): Knowledge search via bot, article sharing, gap reporting
- Analytics (Google Analytics, internal tools): Usage tracking, popular articles, gap identification
- CRM (Salesforce, HubSpot): Account-specific knowledge, customer context
- Documentation (GitBook, ReadMe, Document360): Product documentation as knowledge source
- Project Management (Jira, Linear): Known issues tracking, release notes integration
- AI/Search (Algolia, Elasticsearch): Intelligent search, article recommendations
Edge Cases
- Tribal knowledge not documented: Expert agents know things not in KB
- Knowledge mining: Interview top agents; document their processes
- Incentive: Recognize and reward knowledge contributors
- Requirement: New agents must contribute 2 articles during onboarding
- Exit process: Departing agent documents key knowledge before leaving
- Audit: Quarterly review of "knowledge gaps" reported by agents
- Conflicting information across articles: Two articles say different things
- Single source of truth: Designate one canonical article per topic
- Cross-linking: All related articles link to canonical version
- Review: Content owner resolves conflicts during review cycle
- Flag system: Agents can flag "conflicting information" for review
- Versioning: Track when information changed; explain why
- Knowledge base becomes too large: 2,000+ articles, hard to navigate
- Taxonomy review: Quarterly structure review; merge redundant articles
- Search optimization: Better search > perfect organization
- Tiered access: Role-based visibility (tier-1 vs tier-2 knowledge)
- Archiving: Retire outdated articles; keep in archive for reference
- Curation: Content owners maintain "essential reading" lists per role
- Agents don't use knowledge base: KB exists but agents don't reference it
- Training: Show agents HOW to find and use knowledge during onboarding
- Integration: Knowledge suggestions appear automatically in ticket view
- Culture: Managers reference KB in team meetings; "Check the KB first"
- Quality: Agents won't use poor-quality KB; fix accuracy and completeness
- Measurement: Track KB usage per agent; coach low users
- Product changes faster than documentation: KB always behind product releases
- Change management: Product team notifies knowledge team of changes 2 weeks before release
- Release checklist: "Update KB articles" is part of release process
- Temporary notes: "Coming in v3.5" notes added to affected articles
- Known issues: Maintain running known issues page (updated weekly)
- Agent feedback: "This article doesn't match what I see" → triggers review
- Knowledge for specialized roles: Tier-2/Engineering knowledge not relevant to tier-1
- Tagging: Tag articles by audience (tier-1, tier-2, engineering, billing)
- Views: Create role-specific homepages showing relevant articles
- Search filtering: Filter results by audience tag
- Escalation handoff: Tier-2 knowledge available when ticket escalated
- Progressive disclosure: Show basic info first, advanced info on demand