Support Ticket

Customer support requests are handled verbally or through informal channels — a Slack DM, a text message to the CSM, a passing comment during a call. No support ticket is created. There is no record that the issue was raised, who raised it, or whether it was resolved. 'Did we fix that thing for Acme?' depends on whether anyone remembers the conversation.

None — AI cannot perform any support analysis, routing, or deflection because no support ticket records exist in any system.

Support tickets are created in a helpdesk tool but with minimal structure. A ticket has a subject line, a free-text description, and an assignee. There are no required categories, no severity levels, no product area tags. Finding all support tickets related to the reporting module means keyword-searching 'report' across thousands of free-text descriptions and hoping the customer used that word.

AI could scan support ticket descriptions for keyword patterns, but cannot reliably categorize, route, or prioritize tickets because the free-text format lacks the consistent structure needed for automated triage.

Support tickets have structured fields — category (bug, question, feature request), priority (P1-P4), product area, customer account, and assigned agent. Agents are expected to classify tickets at intake. Managers can filter 'show me all P1 bugs in the API module this month' and get a reliable answer. But the support ticket record is self-contained — it does not link to the customer's usage context, health score, or contract value.

AI can generate support ticket analytics by category, priority, and product area. Can route tickets based on classification. Cannot prioritize by business impact because support tickets lack links to customer account value, health status, or contract renewal proximity.

Support tickets are comprehensive records linked to customer accounts, subscription tiers, health scores, and product usage context. An agent opening a support ticket immediately sees the customer's ARR, health trajectory, recent feature usage, and previous ticket history. A support manager can query 'show me all open P1 tickets from enterprise accounts with health scores below 60 and renewals within 90 days' and get a precise, contextualized answer.

AI can prioritize support tickets by business impact, predict escalation likelihood, and recommend responses based on similar resolved tickets with full customer context. Cannot yet auto-resolve tickets because resolution knowledge is not structured for machine consumption.

The support ticket is a formal entity in a customer service ontology. Each ticket has validated relationships to the customer account, product components, knowledge base articles, root cause taxonomy entries, and resolution playbooks. Classification is machine-readable. An AI agent can ask 'which recurring support ticket patterns for the billing module have no corresponding knowledge base article and affect accounts with ARR above $50K?' and get a structured answer.

AI can autonomously triage, route, and draft responses for support tickets using the structured ontology. Auto-resolution of known issue patterns, intelligent escalation, and proactive ticket deflection operate through ontology-driven reasoning.

Support tickets are living, self-enriching entities. Every incoming request automatically populates with the customer's current session context, recent product usage patterns, similar resolved tickets, applicable knowledge base articles, and real-time system status. The support ticket generates its own context from the customer event stream rather than relying on the customer to describe what happened.

Can autonomously handle the full support ticket lifecycle — from context-aware intake through intelligent routing, AI-assisted resolution, customer satisfaction measurement, and knowledge base contribution — all driven by the self-enriching ticket entity.

Ceiling of the CMC framework for this dimension.