Guided Troubleshooting

Guided Troubleshooting Workflows

Walk customers through diagnostic steps to resolve issues without agent intervention.

Workflow

Troubleshooting Workflow Design

Trigger: New product issue pattern; recurring support category; quarterly workflow review:

1. Issue identification: Analyze support tickets to identify top troubleshooting categories (top 10 issue types representing 60–70% of tickets); categorize by product area and complexity.
2. Root cause mapping: For each issue, map all known causes (typically 5–15 per issue type); rank by frequency (%); determine diagnostic questions that isolate each cause.
3. Decision tree design: Build branching logic — start with broad question; narrow down through successive yes/no questions; each leaf node = specific fix.
4. Fix scripting: For each leaf node, write clear fix steps (numbered, with screenshots/GIFs); include "Did this work?" checkpoint; add escalation path if fix fails.
5. Auto-data collection: Integrate with product APIs to auto-collect diagnostics (account status, error logs, API health) — reduces manual steps by 30–50%.
6. QA testing: Test with 50+ real customer scenarios; validate path efficiency (average steps to resolution ≤ 6); verify 100% of known causes are covered.
7. Launch and monitor: Deploy to self-service portal and chatbot; track resolution rate (target 55–70%); monitor drop-off points; iterate monthly.
8. Continuous improvement: Add new branches from escalated tickets; retire low-usage paths; optimize question order based on analytics.

Troubleshooting Workflow Template

TROUBLESHOOTING WORKFLOW — LOGIN ISSUES (EXAMPLE)
===================================================

Issue: "I can't log in to my account"

Step 1: Can you reach the login page?
  Yes → Go to Step 2
  No → Check internet connection; verify URL (portal.company.com); clear browser cache; try incognito mode
       Did this work? Yes → RESOLVED | No → Step 1b

Step 1b: What error do you see on the login page?
  "Page not found" → Verify correct URL; contact support if persistent
  "Connection timeout" → Check firewall settings; try different network
  "500 Internal Error" → Our system may be down; check status page; wait 15 min and retry
  Other → ESCALATE (collect error screenshot)

Step 2: What happens when you enter your credentials?
  "Password incorrect" → Step 3
  "Account locked" → Step 5
  "Error: [specific code]" → Step 6
  "Loads but nothing happens" → Clear cookies; disable browser extensions; try different browser
  "Account not found" → Verify email address; check for typos; try account recovery

Step 3: Have you tried password reset?
  Yes → Step 4
  No → Click "Forgot Password"; enter email; check inbox/spam for reset link; click link; set new password
       Did this work? Yes → RESOLVED | No → Step 4

Step 4: Is the reset email arriving?
  Yes → Click link within 24 hours; check browser pop-up blockers
  No → Check spam/junk folder; add [email protected] to contacts; try different email address
       If email on file is wrong → Update email via account recovery form
       Did this work? Yes → RESOLVED | No → ESCALATE

Step 5: Account locked — unlock process
  Click "Unlock Account" below login form
  Enter verification code sent to email/phone
  Set new password
  Did this work? Yes → RESOLVED | No → ESCALATE (security review needed)

Step 6: Check specific error code
  Error 1001: Session expired → Log out completely; clear cookies; log back in
  Error 1002: Invalid credentials → Verify username (not email); reset password
  Error 1003: Account suspended → Contact billing (payment issue); call support line
  Error 1004: MFA required → Open authenticator app; enter 6-digit code
  Other error → ESCALATE (collect error code + screenshot)

Escalation: If all steps fail, auto-create ticket with:
  - All steps attempted (with customer answers)
  - Browser/OS info (auto-detected)
  - Account ID and login email
  - Time of last attempt
  - Route to: Technical Support Queue (Priority: Medium)

Workflow Analytics Dashboard

TROUBLESHOU BING WORKFLOW PERFORMANCE METRICS
================================================

Workflow: Login Issues
Last updated: Monthly

Volume Metrics:
  Total sessions: 1,250/month
  Self-resolved: 68% (850 sessions)
  Escalated to agent: 32% (400 sessions)
  Abandoned mid-flow: 12% (150 sessions)

Efficiency Metrics:
  Average steps to resolution: 4.2 (target: ≤ 6)
  Average time to resolution: 3.5 minutes (target: ≤ 5)
  Most efficient path: Step 1 → Step 2 → Step 3 → Reset (62% of resolutions)
  Least efficient path: Step 1 → Step 1b → Step 2 → Step 6 → Escalate (8% of sessions)

Drop-off Analysis:
  Step 1 drop-off: 5% (low — question is clear)
  Step 2 drop-off: 8% (acceptable)
  Step 3 drop-off: 18% (HIGH — password reset step too complex; simplify)
  Step 4 drop-off: 15% (HIGH — email troubleshooting needs better guidance)

Optimization Recommendations:
  1. Simplify Step 3: Add direct reset link instead of instructions
  2. Add Step 1.5: Quick "Are you using the right email?" check
  3. Add video guide for Step 4 (email troubleshooting)
  4. New branch: "Forgot which email I used?" → account recovery by phone number

ROI:
  Tickets deflected: 850/month (× 5 min agent time = 70.8 agent-hours saved/month)
  Agent cost savings: $2,124/month (at $30/hour fully-loaded)
  Annual savings: $25,488
  Customer satisfaction: 4.3/5.0 (resolved without waiting for agent)

Edge Cases

• Ambiguous symptoms (customer describes issue in vague terms):
• Example: "My account isn't working"
• Strategy: Start with broad triage questions; use visual aids (screenshots of expected vs. actual); ask for screenshots
• Question design: Use "What are you trying to do?" instead of "What error do you see?"
• Fallback: If unable to narrow after 3 questions, escalate with all attempted paths

• Multi-component issues (root cause spans multiple systems):
• Example: Integration failure — could be API, network, or configuration
• Strategy: Parallel diagnostic branches; check each component independently; present combined results
• Auto-checks: API status endpoint; network connectivity test; configuration validation tool
• Complexity limit: If 3+ components involved, escalate to engineer with full diagnostic report

• Outdated troubleshooting paths (product changed but workflow not updated):
• Detection: Resolution rate drops below baseline; customer feedback "this didn't work" increases
• Prevention: Tie workflow updates to product release notes; automated flag when KB article referenced is updated
• SLA: Update workflows within 48 hours of product change affecting covered issues
• Review: Monthly review of all active workflows against current product version

• Frustrated customer experience (customer already tried basic steps before reaching bot):
• Detection: Customer says "I already tried that" or sentiment analysis shows frustration
• Adaptation: Skip common steps; ask "What have you already tried?" (checkbox selection); jump to advanced steps
• Proactive: At start, offer "Already tried the basics?" shortcut to advanced troubleshooting
• Empathy: When frustration detected, add empathetic language: "I understand this is frustrating. Let's try something different."

• Regulatory/compliance troubleshooting (security-related issues):
• Example: Suspected unauthorized access, data breach
• Rule: NEVER guide through security investigation; escalate immediately to security team
• Immediate action: Auto-lock account (if confirmed breach); notify customer; create P1 ticket to security
• Documentation: Log all troubleshooting steps for forensic analysis
• Communication: Use approved incident communication templates only

Integration Points

• Help desk platforms: Zendesk, Freshdesk — ticket auto-creation on escalation; conversation history transfer
• Product APIs: Custom endpoints for account status, error logs, API health checks
• Chatbot platforms: Intercom, Drift, Dialogflow — embed troubleshooting flows in conversational UI
• Knowledge base: Zendesk Guide, Confluence — link troubleshooting steps to detailed articles
• Customer data: Salesforce, HubSpot — account tier, plan details, usage history
• Analytics: Mixpanel, Amplitude — step-by-step funnel tracking; drop-off analysis
• Screen recording: Loom, Lookback — customers can record screen for context
• Error tracking: Sentry, Bugsnag — correlate customer errors with known issues
• Status page: Atlassian Statuspage — auto-reference system status during troubleshooting
• Data warehouse: Snowflake — historical troubleshooting data for pattern analysis

Disclaimer: All rights reserved by Circulos AI. These skills are specifically designed for Claude Code, Claude Cowork, Codex, and OpenClaw. When using or referencing any skill, please provide proper attribution to Circulos AI.