Customer Service Quality Auditor

Problem it addresses

Manual QA of customer-service conversations often relies on impression-based scoring: the same phrase may be acceptable in one context and inflammatory in another. This skill turns chat logs into an auditable scoring structure, requiring each dimension to be rated from 1 to 5 with quoted source text as evidence for the score.

How it works

• Input parsing: supports json, csv, txt, or pasted text, and normalizes agent/customer roles, message content, and optional timestamps.
• Dimension scoring: rates response speed, politeness, issue understanding, solution quality, communication skill, proactive service, compliance, and customer emotion management using weighted criteria.
• Session grading: maps total scores to A/B/C/D tiers for benchmark, pass, improve, and critical cases.
• Team analytics: produces per-agent averages, dimension distributions, and common weaknesses, including dimensions where 30% or more agents score below 3.
• Improvement advice: follows “problem → root cause → action → reference script” for individual and team recommendations.

Boundaries

It is best used as an AI-assisted first pass for QA review and training material prep, not as a final human-free judgment. Without timestamps, response-time scoring defaults to 3; reports must mask customer names, phone numbers, and other sensitive fields and keep a manual-review disclaimer. It should also separate agent skill gaps from system or process defects.