QualityOS Docs
Docs/Core Features/Call QA

Call QA

Automatically score customer calls against 8 weighted parameters with AI-generated evidence and coaching notes.

Input methods

Paste transcript

Copy a call transcript in any readable format (AGENT / CUSTOMER turns work best) and paste into the text area. No file upload needed.

Upload audio

Upload an MP3 or WAV file. QualityOS sends it to Groq Whisper for transcription in memory, then runs the full analysis. Audio is never stored.

Scoring parameters

Each parameter is scored 0–10 and combined into an overall score /100 using the weights below.

ParameterWeight

Greeting

Opening professionalism, name introduction, tone set in the first 30 seconds.

8%

Empathy

Acknowledgement of the customer's situation and emotional state throughout the call.

15%

Active Listening

Confirmation of understanding, no interruptions, follow-up questions that show engagement.

10%

Tone & Professionalism

Language register, avoidance of fillers, composure under difficult conditions.

12%

Resolution Accuracy

Correctness of the answer or solution provided. KB-grounded when a document is uploaded.

20%

Policy Compliance

Adherence to your SOPs and company policies. KB-grounded when a document is uploaded.

15%

Sentiment Arc

How customer sentiment changed from start to end of the call.

10%

Closing Quality

Confirmation of resolution, next steps communicated, professional sign-off.

10%

Verdicts

PASS

Score ≥ 85. Agent met quality standards.

COACH

Score 70–84. Coaching opportunity identified.

FLAG

Score < 70. Escalation or intervention needed.

Evidence quotes

For each parameter, QualityOS pulls verbatim excerpts from the transcript to justify the score. This makes every result auditable and removes guesswork from coaching conversations.

Knowledge Base grounding

When you have documents uploaded in Settings → Knowledge Base, QualityOS retrieves the most relevant chunks from your SOPs and policies using BM25 + Groq reranking, then injects them into the analysis prompt.

Impact by parameter

Resolution Accuracy — scored against your actual answer templates and escalation policies

Policy Compliance — verified against your specific company policies, not generic QA rubrics

Other parameters use best-practice defaults regardless of KB presence

Coaching brief

Every scorecard includes a three-part coaching brief ready to paste into a 1:1:

  • Positive callout — what the agent did well, with a transcript reference
  • Top miss — the single highest-impact improvement area
  • Coaching note — a suggested talking point for the manager's 1:1 conversation