Resolution Methodology

1. What qualifies as an L1 ticket

TicketResolve AI is designed for Level 1 (L1) support tickets — requests that are repetitive, low-complexity, and answerable from existing knowledge without requiring human judgment, account-level investigation, or escalation to engineering.

Typical L1 ticket categories include:

• Password resets and account access issues
• Billing and subscription queries covered by published policy
• How-to questions answered in the knowledge base or documentation
• Order status and tracking queries (where integration data is available)
• Known bug workarounds documented in resolved ticket history
• Third-party integration questions covered by approved external sources

Tickets are not classified as L1 when they involve account-specific data requiring human lookup, legal or compliance sensitivity, emotional distress signals, or ambiguity that falls below the configured confidence threshold.

2. What counts as an automated resolution

A ticket is counted as auto-resolved when all of the following conditions are met:

• TicketResolve AI generates a response from one or more approved knowledge sources
• The response confidence score meets or exceeds the customer's configured threshold
• The response is sent to the customer without human review or modification
• The customer does not reopen the ticket within 48 hours of the automated response

Tickets that are escalated to a human agent — for any reason — are never counted as auto-resolved, even if the agent subsequently closes the ticket without further action.

3. Response, deflection and complete resolution — the difference

These three terms are often conflated in AI support marketing. TicketResolve AI uses precise definitions:

4. How accuracy is calculated

Accuracy will be measured as the proportion of auto-resolved tickets that receive no negative signal within a defined evaluation window. Negative signals include:

• Customer reopens the ticket within 48 hours
• Customer explicitly marks the response as unhelpful
• Customer submits a follow-up ticket on the same issue within 7 days
• Human agent overrides the automated response after review

Accuracy will be reported separately for each knowledge source combination (knowledge base only; knowledge base + ticket history; all five intelligence sources) and for each ticket category.

5. How reopened tickets are treated

Any ticket reopened within 48 hours of an automated response is reclassified as unresolved. It is removed from the auto-resolution count and routed to a human agent. The original automated response is flagged for quality review.

Tickets reopened after 48 hours are treated as new tickets and processed independently.

6. How escalations are measured

The escalation rate is the proportion of tickets that are routed to a human agent rather than auto-resolved. Escalations occur when:

• The confidence score falls below the customer's configured threshold
• The ticket matches a category the customer has designated as always-human
• Sentiment analysis detects emotional distress or urgency signals
• The ticket contains keywords on the customer's escalation list
• No relevant content is found across all approved knowledge sources

Escalated tickets are never counted as auto-resolutions. The escalation rate is reported in the customer dashboard alongside the auto-resolution rate.

7. Evaluation sample size and measurement period

When aggregate benchmarks are published, they will include: total tickets evaluated, number of customers in the sample, date range, breakdown by ticket category, and breakdown by knowledge source configuration. Benchmarks will be updated quarterly.

8. Confidence thresholds

Every candidate response is assigned a confidence score before it is sent. The score reflects how well the retrieved source content matches the ticket intent, the recency and authority of the source, and the consistency of evidence across multiple sources.

Thresholds can be set globally or per ticket category. A higher threshold reduces auto-resolution rate but increases the reliability of responses that are sent.

9. Human review process

Escalated tickets arrive in the human agent queue with a structured handoff package containing:

• Ticket summary and intent classification
• Sources checked and relevance scores
• Draft response (if one was generated before confidence fell below threshold)
• Escalation reason (low confidence, category rule, sentiment signal, or no match)
• Confidence score at the point of escalation

Agents can accept the draft response, modify it, or write a new response from scratch. All agent actions on escalated tickets are logged.

10. Known limitations

• Knowledge base quality dependency. Auto-resolution rates are directly correlated with the completeness and accuracy of the customer's knowledge base. Teams with sparse or outdated documentation will see lower initial rates.
• Ticket history volume. The resolved-ticket learning source requires sufficient historical data to be effective. New deployments with limited ticket history rely more heavily on the knowledge base.
• Web intelligence scope. Governed web intelligence is limited to customer-approved domains: TR-AI first reads matching pages of those domains directly (live site search: sitemaps, help-center search, status pages), and uses a paid web search of the same allowlist only when that cannot answer well enough. Information on non-approved domains is not accessed, which may result in escalation for tickets requiring that information.
• Language support. Current performance benchmarks apply to English-language tickets. Performance in other languages has not been independently validated.
• Novel issues. TicketResolve AI cannot resolve tickets about issues that have never appeared in the knowledge base, ticket history, or approved external sources. These are escalated.
• No certifications yet. TicketResolve AI does not currently hold SOC 2 Type II, ISO 27001, HIPAA, or GDPR certification. These are on the product roadmap. Do not rely on this product for regulated data environments without independent legal review.

11. Claims removed or qualified

The following claims have been removed from the TicketResolve AI website pending verified evidence:

• "94% accuracy" — removed; accuracy measurement framework is planned, not yet validated
• "78-80% average auto-resolution rate" — replaced with "up to 80%" with qualification
• "847,293 tickets processed" — removed; aggregate volume figure not independently verified
• "SOC 2 Type II" — removed from pricing comparison; certification not yet held