AI Agent Escalation Breakdown for Tickets

Quack’s AI Ticket Agents are designed to resolve support tickets end to end. In certain situations, Quack may escalate a ticket: either as expected per your setup or due to system or AI limitations. Understanding escalation types helps you fine-tune your setup and continuously improve self-service coverage.

Escalation Overview

🟢 Handled Successfully by Quack

• Fully Resolved by Quack: Ticket was answered and marked as resolved

• Customer Did Not Reply Back: Quack responded, but the customer did not provide any further input

🟡 Expected Escalations (Per Configuration)

• Topic Not Released: Subject not enabled for Quack to handle

• Out of Policy: Request restricted due to business or legal policy

• Strict Mode Triggered: Escalation due to fallback being disabled

🔴 Unexpected Escalations (AI Limitations)

• Knowledge Gap: Missing or insufficient training data

• Low Confidence: Quack was not confident enough to respond

• System Error: Escalation caused by a technical malfunction

Details by Escalation Type

🟢 Handled Successfully by Quack

Fully Resolved by Quack
What it means: Quack provided an answer and resolved the ticket without human involvement
Action: No action needed. Review responses periodically for quality assurance

Customer Did Not Reply Back
What it means: Quack responded, but the customer did not add further comments or questions
Action: No action needed unless a recurring pattern suggests customer confusion or dissatisfaction

🟡 Expected Escalations (Per Configuration)

Topic Not Released
What it means: The requested subject is not enabled for Quack in your configuration so Quack skipped the ticket
Action: Enable additional topics if appropriate

Out of Policy
What it means: The request is restricted due to business, legal, or compliance rules
Action: Update briefs if policies change

💡 Pro Tip: Configure briefs to collect key details before escalation when needed

Strict Mode Triggered
What it means: Strict Mode settings blocked fallback answers
Action: Add training if appropriate, or adjust Strict Mode configuration

🔴 Unexpected Escalations (AI Limitations)

Knowledge Gap
What it means: Quack could not answer due to missing or insufficient training
Action: Expand training data, add or update articles, improve structure

Low Confidence
What it means: Quack was not sufficiently confident in its answer, or the customer’s phrasing was too ambiguous
Action: Add ambiguous examples to training and refine phrasing

System Error
What it means: A technical error caused escalation, such as API timeout or backend issue
Action: Contact the Quack team for investigation

Quick Reference: Who Owns What