--- name: escalation-routing description: Automatically escalate complex support issues to appropriate specialists and leadership based on issue type, customer tier, severity, sentiment spikes, SLA risk, and expertise requirements. Use when setting up escalation workflows, configuring escalation triggers, managing on-call rotations, handling VIP escalations, or optimizing escalation processes. Triggers on phrases like "escalation routing", "auto-escalation", "escalation workflow", "on-call rotation", "ticket escalation", "escalation trigger", "escalation SLA", "L2 escalation". --- # Intelligent Escalation Routing Automatically route complex or critical support issues to the appropriate specialists, leadership, or teams based on multi-factor analysis including issue type, customer tier, severity, and urgency. ## Workflow ### 1. Escalation Trigger Detection 1. **Escalation trigger types**: ``` ESCALATION TRIGGER MATRIX ========================= CUSTOMER-INITIATED TRIGGERS: Explicit escalation request: "manager", "supervisor", "escalate", "speak to someone senior" Complaint language: "unacceptable", "terrible service", "waste of time" Repeated contacts: 3+ tickets on same issue within 48 hours VIP customer flag: Enterprise/strategic account with any critical issue AGENT-INITIATED TRIGGERS: Agent marks "needs escalation" (complexity beyond their scope) Agent unable to resolve after 30 minutes of investigation Agent identifies potential product bug Agent detects customer dissatisfaction rising despite efforts SYSTEM-AUTOMATIC TRIGGERS: SLA breach imminent (95% of SLA time elapsed with no resolution) Priority P0/P1 ticket created (security, data loss, system outage) Sentiment score drops below -0.7 (extremely frustrated/angry) Legal/security keywords detected: "lawsuit", "breach", "data leak" Revenue impact >$10,000 (enterprise customer, business critical) Mass ticket pattern: 5+ similar tickets in 1 hour (potential outage) CONDITION-BASED TRIGGERS: Ticket age exceeds threshold for priority level: P0: 30 minutes without response P1: 2 hours without response P2: 8 hours without resolution P3: 24 hours without resolution Failed self-service attempts: Customer tried chatbot + KB + still unresolved Customer tier degradation: Account health score dropped 20+ points ``` 2. **Escalation path determination**: - **Technical issues** → L2 Technical Specialist → Engineering (if bug) - **Billing disputes** → Billing Specialist → Finance Manager (>$1,000) - **Account security** → Security Team (immediate, bypass normal queue) - **VIP/Executive** → Dedicated Senior Agent → Support Manager → CSM - **Legal/Compliance** → Legal Team (immediate, auto-hold on responses) - **Product feedback** → Product Manager (via backlog system) - **System outage** → On-call Engineering → Incident Commander ### 2. Escalation Execution and Notification 1. **Multi-channel notification system**: ``` NOTIFICATION CADENCE BY SEVERITY: ================================= P0 (Critical — System Down / Data Loss): Immediate (T+0): • PagerDuty alert to on-call engineer (phone call + SMS) • Slack notification to #incidents channel • Email to support manager + engineering manager • Status page update (if affecting multiple customers) T+5 min: • If no acknowledgment → escalate to next on-call + VP • War room created (Slack + Zoom bridge) T+15 min: • If no resolution → executive notification • Proactive customer communication initiated T+30 min: • If no resolution → VP Engineering + VP Support notified P1 (High — Major Feature Broken / VIP Customer): Immediate (T+0): • Slack notification to #escalations channel + @on-call • Email to support manager T+15 min: • If no acknowledgment → phone call to on-call specialist T+30 min: • If no resolution → support manager takes ownership P2 (Medium — Complex Issue Requiring Specialist): Immediate (T+0): • In-app notification to specialist queue • Slack message to relevant team channel T+1 hour: • If no pick-up → reassign from pool + manager notification T+4 hours: • If unresolved → manager assignment P3 (Low — Routine Escalation): Immediate (T+0): • Ticket moved to specialist queue • In-app notification to assigned specialist T+24 hours: • If unresolved → manager review ``` 2. **Context preservation**: - Full ticket history included in escalation - Agent's troubleshooting notes and steps already attempted - Customer communication preferences and sentiment - Account details and health score - Time spent on ticket by each handler - Suggested resolution approach from AI analysis ### 3. Escalation Resolution and Feedback Loop 1. **Resolution workflow**: - Specialist investigates and resolves issue - Resolution documented in ticket - Ticket returned to original agent for customer communication - Original agent informs customer of resolution - Customer CSAT survey sent - Resolution quality reviewed (was the escalation necessary?) 2. **Post-escalation analysis**: - Was escalation appropriate? (yes/no — agent feedback) - Could it have been prevented? (training gap, tool gap, process gap) - Was resolution quality acceptable? (CSAT, reopen rate) - Time-to-resolution vs SLA - Agent learning: What can L1 do better next time? 3. **Continuous improvement**: - Monthly escalation review meeting (support manager + team leads) - Escalation rate tracking by agent (identify training needs) - Escalation reason analysis (identify systemic issues) - KB article creation from escalation resolutions (prevent future escalations) ## Templates & Frameworks ### Escalation Dashboard ``` ESCALATION MANAGEMENT DASHBOARD — January 2026 ================================================ ESCALATION VOLUME: Total escalations this month: 312 (↑ 8% from December) By severity: P0 (Critical): 8 (2.6%) P1 (High): 34 (10.9%) P2 (Medium): 156 (50.0%) P3 (Low): 114 (36.5%) By trigger type: Agent-initiated: 145 (46.5%) System-automatic: 102 (32.7%) Customer-requested: 42 (13.5%) VIP routing: 23 (7.4%) ESCALATION PERFORMANCE: Metric | Current | Target | Status --------------------------------|---------|----------|-------- Avg acknowledgment time (P0) | 3.2 min | <5 min | ✓ Avg acknowledgment time (P1) | 8.7 min | <15 min | ✓ Avg acknowledgment time (P2) | 23 min | <30 min | ✓ Avg resolution time (P0) | 2.1 hrs | <4 hrs | ✓ Avg resolution time (P1) | 6.4 hrs | <8 hrs | ✓ Avg resolution time (P2) | 18 hrs | <24 hrs | ⚠ close Escalation resolution rate | 93.4% | >90% | ✓ Customer CSAT (escalated) | 4.1/5.0 | >4.0 | ✓ Unnecessary escalations | 28 (9.0%)| <10% | ⚠ close TREND: Resolution times ↓ 12% from last month (improving) CURRENT ESCALATIONS (Active): Priority | Ticket ID | Customer | Issue | Escalated To | Time Open ---------|-----------|--------------|----------------------------|--------------|---------- P0 | TKT-99234 | Globex Corp | API outage (all endpoints) | On-call Eng | 23 min ⚠ P1 | TKT-99187 | Initech | SSO broken for all users | L2 Specialist| 1.2 hrs P1 | TKT-99156 | ACME Inc | Data export failing | L2 Specialist| 45 min P2 | TKT-99098 | Wayne Ent | Custom report error | L2 Specialist| 2.5 hrs P2 | TKT-99045 | Stark Ind | Webhook delivery failure | L2 Specialist| 1.8 hrs P2 | TKT-98992 | Upton Inc | Billing discrepancy | Billing Spec | 3.2 hrs Total active escalations: 6 (within capacity: 10) ESCALATION ROUTE PERFORMANCE: Route | Escalations | Resolution Rate | Avg Resolution Time | CSAT -------------------------|-------------|-----------------|---------------------|------ L2 Technical Specialist | 124 | 94.4% | 14.2 hrs | 4.2/5.0 Engineering (bugs) | 45 | 91.1% | 36.8 hrs | 4.0/5.0 Billing Specialist | 67 | 97.0% | 4.5 hrs | 4.4/5.0 Security Team | 12 | 100% | 2.1 hrs | 4.3/5.0 Support Manager | 34 | 94.1% | 8.7 hrs | 4.1/5.0 Legal Team | 8 | 100% | 24.5 hrs | 3.8/5.0 On-call Engineering (P0) | 22 | 100% | 1.8 hrs | 4.5/5.0 AGENT ESCALATION RATES: Agent | Escalation Rate | Team Avg | Status | Coaching Needed ---------------|-----------------|----------|-----------|---------------- Sarah M. | 6.2% | 11.2% | ✓ Below | No Tom R. | 8.1% | 11.2% | ✓ Below | No James T. | 18.4% | 11.2% | ⚠ Above | Yes — training Kevin W. | 22.3% | 11.2% | ⚠ High | Yes — urgent Maria L. | 9.5% | 11.2% | ✓ Normal | No Top 3 agents for unnecessary escalations: Kevin W.: 12 unnecessary (55% of his escalations) → training needed James T.: 8 unnecessary (43% of his escalations) → coaching needed Rachel H.: 5 unnecessary (40% of her escalations) → review needed KNOWLEDGE CAPTURE: Escalation resolutions added to KB: 34 articles Agent training materials from escalations: 8 guides Process improvements identified: 5 TOP ESCALATION REASONS (Prevention Opportunities): 1. API troubleshooting (89 escalations) → Create L1 API guide (in progress) 2. SSO configuration (56 escalations) → Add decision tree to KB (planned) 3. Complex billing disputes (45 escalations) → Agent billing training (scheduled) 4. Custom report errors (34 escalations) → Product documentation update (requested) 5. Integration webhook failures (28 escalations) → Troubleshooting checklist (in progress) ON-CALL ROTATION: Current on-call engineer: Dr. Chen (until Jan 17, 17:00 UTC) Next on-call: Sarah Kim (Jan 17, 17:00 UTC) Escalation acknowledgment rate (on-call): 98.7% Average acknowledgment time: 3.2 minutes ON-CALL HEALTH: Pages received this rotation: 12 Acknowledged within 5 min: 12 (100%) Average acknowledgment time: 3.2 minutes False alerts: 2 (16.7%) ``` ### Escalation Playbook ``` ESCALATION DECISION TREE ========================= TICKET RECEIVED → Is escalation needed? STEP 1: Check for AUTO-ESCALATION triggers (immediate): • P0 priority? → YES → Escalate to on-call engineering immediately • Security/legal keywords? → YES → Escalate to security/legal team • VIP + critical issue? → YES → Escalate to senior agent + support manager • SLA at 95%+ with no resolution? → YES → Auto-escalate to next level STEP 2: Check for AGENT ESCALATION: • Can you resolve with available tools and knowledge? YES → Continue handling NO → Proceed to Step 3 STEP 3: Identify ESCALATION TYPE: • Technical depth needed? → Route to L2 Technical Specialist • Billing complexity? → Route to Billing Specialist • Potential product bug? → Route to Engineering (via Jira integration) • Customer demanding manager? → Route to Support Manager • Account/contract issue? → Route to CSM STEP 4: Prepare ESCALATION PACKAGE: ✓ Document all troubleshooting steps attempted ✓ Attach relevant screenshots, logs, error messages ✓ Note customer sentiment and communication style ✓ Include customer account context (tier, history, health score) ✓ Add clear escalation reason: "Need help because..." ✓ Set appropriate priority level (don't auto-set to P0) STEP 5: After RESOLUTION: ✓ Receive resolution details from specialist ✓ Communicate resolution to customer (in your voice) ✓ Verify customer satisfaction ✓ Document what you learned (personal knowledge capture) ✓ Flag if KB article needed (prevent future escalation) WHAT NOT TO DO: ✗ Don't escalate without attempting reasonable troubleshooting first ✗ Don't escalate billing questions above your refund limit ✗ Don't set P0 priority unless system is truly down ✗ Don't escalate just because customer is angry (handle with empathy first) ✗ Don't forget to document troubleshooting steps before escalating ``` ## Integration Points - **Help desk** (Zendesk, Intercom, Freshdesk): Ticket routing, priority management, SLA tracking - **Paging systems** (PagerDuty, Opsgenie): On-call rotation, critical alerts, escalation chains - **Communication** (Slack, Teams, SMS): Multi-channel notifications, team collaboration - **CRM** (Salesforce, HubSpot): VIP customer identification, account tier, CSM contact - **Engineering systems** (Jira, Linear, GitHub): Bug report creation, engineering ticket tracking - **Knowledge base**: Internal KB for escalation procedures, resolution capture - **Analytics** (Tableau, Power BI): Escalation metrics, trend analysis, performance reporting ## Edge Cases - **On-call specialist unavailable during P0 incident**: Primary on-call doesn't acknowledge within 5 minutes: - Auto-escalate: Secondary on-call paged immediately - Manager notification: Support manager alerted of acknowledgment failure - Escalation chain: VP notified after 15 minutes without acknowledgment - War room: Auto-created with all available team members - Post-incident: Review why primary on-call was unavailable - **Escalation ping-pong between teams**: Ticket escalated back and forth between L2 and Engineering 3+ times: - Circuit breaker: After 2 bounces, requires manager arbitration - Ownership clarity: Manager assigns definitive owner - Root cause: Was initial routing wrong? Was resolution insufficient? - Prevention: Clear escalation guidelines per team, ownership matrix - Tracking: Bounce rate monitored as quality metric - **Agent over-escalating to avoid difficult tickets**: Agent escalates 40% of tickets (vs 11% team avg): - Pattern detection: System flags unusually high escalation rate - Quality review: Sample of agent's escalations reviewed - Coaching: Specific training on confidence, resources available - Support: Mentor pairing with senior agent - Accountability: Escalation rate included in performance review - **Escalation during holiday/after-hours**: P1 ticket escalated when no specialists available: - After-hours on-call: Designated specialist on rotation - Auto-pause: SLA timer paused during off-hours (for non-critical) - Customer communication: "Your issue has been escalated and will be addressed by [time]" - Queue priority: Escalated tickets prioritized at start of next business day - Emergency contact: Critical issues → phone contact for on-call specialist - **Customer requests specific person who is unavailable**: Customer demands to speak with John (top agent) who is on PTO: - Coverage assignment: John's designated backup agent handles the ticket - Transparency: "John is currently unavailable. I'm his designated backup and have full context on your account" - Continuity: John's notes and history transferred to backup - Follow-up: John reviews ticket upon return and contacts customer if needed - Prevention: Backup agent assignment visible to customers in self-service portal ## Output ### Monthly Escalation Report ``` ESCALATION MONTHLY REPORT — January 2026 ========================================== VOLUME SUMMARY: Total escalations: 312 (↑ 8% from December) Escalation rate: 11.2% of total tickets (stable, target: <12%) ✓ RESOLUTION EFFECTIVENESS: Resolved at first escalation: 268 (85.9%) Required re-escalation: 31 (9.9%) Unresolved after escalation: 13 (4.2%) Average resolution time: 12.4 hours (↓ 12% from December) ✓ CUSTOMER IMPACT: CSAT for escalated tickets: 4.1/5.0 (↑ 0.2 from December) Reopened escalated tickets: 18 (5.8%) VIP escalations: 23 (all resolved within SLA) ✓ P0 incidents: 8 (average resolution: 2.1 hours) ✓ PROCESS IMPROVEMENTS: KB articles created from escalations: 34 Agent training guides developed: 8 Unnecessary escalation rate: 9.0% (↓ 2% from December) ✓ Top prevention opportunity: API troubleshooting (89 escalations → guide in progress) RECOMMENDATIONS: 1. Complete L1 API troubleshooting guide (target: reduce 89 escalations by 60%) 2. Address Kevin W.'s high escalation rate (22.3%) — coaching scheduled Jan 22 3. Create SSO configuration decision tree (56 escalations — second highest) 4. Review escalation routing for billing (97% resolution at billing team — correct) 5. Plan Q1 escalation reduction target: 10% (from 11.2% to 10.1%) ```