Audit beginner Works with: Any AI

Support Ticket Reply Quality Scorecard

Rubric scoring Has example

Score replies against a clear rubric.

The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.

[VERIFY] is an editorial flag, not a blank to fill. Wherever the prompt shows [VERIFY], check that item against a reliable source before you use it.

Score five support replies. For each, paste the customer message, the agent reply and the outcome: [TICKET_SAMPLES]. Our policies and tone guidelines: [POLICIES_TONE]. What good looks like to us: [QUALITY_DEFINITION].

Score each reply from 1 to 10 on accuracy, completeness, empathy, clarity, policy compliance and next step. Quote the lines that earn or lose points. Check that the reply answers the question asked, not a nearby one, and that nothing promises what policy does not allow.

Return a scorecard table, three patterns across the tickets, a rewritten version of the weakest reply under 120 words, and coaching notes written kindly for the agent. Remove personal data before sharing.

Also check that the scoring is consistent by scoring one reply twice with a gap between, and say which criteria were hardest to judge from text alone.

Then convert the scorecard into a repeatable quality programme. Define the sampling plan, such as five tickets per agent per week chosen at random, the reviewer rota, and the calibration session in which reviewers score the same tickets and discuss differences. Create the final rubric with anchored descriptions for scores of 3, 6 and 9 on each criterion. Provide a coaching conversation outline that begins with strengths, uses specific quotes, agrees one improvement and sets a time to review. Add rules that keep the scorecard fair, such as not using it alone for pay or discipline, handling disputes, and tracking reviewer bias. Suggest metrics to report monthly at team level, and a way to feed common issues back to product and policy owners. Employment and monitoring rules are [VERIFY] for [JURISDICTION] with HR.

Add a closing note on what to do if two reviewers score the same reply differently, including how to record the disagreement and what it teaches about the rubric.

Close with a short message I could send to the person responsible for the next step, asking for what is needed in a polite and specific way.

Fill in before use

Optional: type your details and the prompt above updates. Nothing is saved or sent.

Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.

Scorecard (invented tickets)
Ticket A (refund): accuracy 9, completeness 8, empathy 6, clarity 8, policy 9, next step 9. Strong; apology came after the policy.
Ticket B (login): accuracy 5 (wrong reset steps), completeness 4, empathy 7, clarity 5, policy 8, next step 3. Quote: "Try again later."
Ticket C (delivery): accuracy 8, completeness 7, empathy 9, clarity 8, policy 5 (promised a date), next step 7.
Patterns: apologies late; promises beyond policy; no next step on technical issues.
Rewrite of Ticket B in 90 words with exact steps and a time for follow-up. Coaching: begin with acknowledgement; end with the next action.

• audit