• Overview
• Per-Question Evaluation Modes
  • What is Per-Question Evaluation Mode?
  • Prerequisites
• Configure Evaluation Mode for a Question
• Evaluation Mode Definitions
  • Manual
  • Keyword Match
  • GenAI
• Evaluator Experience
• AI Rewrite for QM Assist Evaluator
  • What is AI Rewrite?
  • Prerequisites
  • Using AI Rewrite
• Best Practices
• FAQ

Overview

Talkdesk QMA/E now supports:

• Per-question evaluation methods.
• AI-powered question rewriting for QM Assist Evaluator (QMAE).

These features help admins and form editors create more flexible, scalable, and AI-friendly evaluation forms.

With per-question evaluation modes, each question in a QM form can use a different scoring method based on the evaluation use case. With AI Rewrite, users can optimize manually written questions into clearer prompts for GenAI evaluation.

Per-Question Evaluation Modes

What is Per-Question Evaluation Mode?

Per-question evaluation mode allows admins to define how each question in a QM form should be evaluated.

Supported evaluation modes:

• Manual
• Keyword Match
• Generative AI

This gives organizations more flexibility when building evaluation forms. For example:

• Objective compliance checks can use Keyword Match.
• Subjective soft-skill questions can remain Manual.
• Conversational quality questions can use Generative AI.

Different evaluation modes can coexist within the same form.

Prerequisites

• QM Form edit permissions.
• Generative AI mode requires:
  • Talkdesk QM Assist Evaluator (QMAE).
  • Transcript availability for the interaction.

Configure Evaluation Mode for a Question

To configure the evaluation mode:

1. Open Talkdesk QM.
2. Navigate to Form.
3. Edit an existing form or create a new form.
4. Select a question.
5. Locate the Evaluation Mode field.
6. Choose one of the following options:
   • Manual
   • Keyword Match
   • Generative AI
7. Save the form.

Evaluation Mode Definitions

Manual

Questions are scored directly by the evaluator.

Best for:

• Subjective reviews.
• Tone and empathy evaluation.
• Situations requiring human judgment.

Keyword Match

Questions are scored based on configured keywords or phrases detected in the transcript.

Best for:

• Compliance statements.
• Script adherence.
• Verification checks.

GenAI

Questions are evaluated by Talkdesk QM Assist Evaluator using GenAI.

Best for:

• Conversational quality.
• Soft skills.
• Complex interaction analysis.
• Context-aware evaluation.

Note: Non-QMAE customers can only use Manual and Keyword Match modes.

Evaluator Experience

During evaluations, evaluators can see which scoring method is applied to each question.

Depending on the configured mode:

• Manual questions require evaluator input.
• Keyword Match questions are automatically scored based on keyword detection.
• GenAI questions are automatically evaluated using AI-generated scoring and reasoning.

AI Rewrite for QM Assist Evaluator

What is AI Rewrite?

AI Rewrite helps form editors optimize evaluation questions for Generative AI scoring.

The feature transforms manually written questions into clearer, more structured, and AI-friendly prompts.

This helps:

• Improve AI scoring consistency.
• Reduce ambiguous wording.
• Minimize manual prompt engineering.
• Improve evaluation accuracy.

Prerequisites

• Talkdesk QM Assist Evaluator enabled.
• QM Form edit permissions.
• The generative AI evaluation mode is enabled for the question.

Using AI Rewrite

To rewrite a question:

1. Open Talkdesk QM.
2. Navigate to Form Builder.
3. Open or create a form.
4. Select a question configured for Generative AI evaluation.
5. Enter your draft question.
6. Click Rewrite.
7. Review the AI-generated version.
8. Choose one of the following actions:
   • Confirm to keep the rewritten version.
   • Undo to restore the original question.
9. Save the form.

Best Practices

For the best Generative AI evaluation results:

• Use clear and objective wording.
• Focus on observable behaviors.
• Avoid overly broad or vague questions.
• Include measurable expectations when possible.
• Use AI Rewrite to refine short or informal prompts.

FAQ (Frequently Asked Questions)

Can different questions use different evaluation methods in the same form?

Yes. A single form can contain Manual, Keyword Match, and Generative AI questions together.

Can non-QMAE customers use per-question evaluation modes?

Yes. Non-QMAE customers can use Manual and Keyword Match modes.

Does AI Rewrite automatically save changes?

No. Users must confirm the rewritten content and save the form.

Can evaluators change the evaluation mode during the scoring process?

No. Evaluation modes are configured by admins or form editors in the Form Builder.

Does Generative AI evaluation require transcripts?

Yes. GenAI evaluation requires transcript availability for the interaction.