Library › Customer Support & Success AI › Customer support teams
Score a support reply on a four-point QA scorecard
- For Customer support teams
- You practice Catching mistakes and made-up facts
- Time about 5 minutes
- Level Advanced
About this exercise
Score the agent reply on the four criteria below. Output a table with the header "Criterion | Score | Evidence", one row per criterion, Score 1 or 0, Evidence a direct quote of fewer than 8 words (or "none" if it fails). Then a last line "Total: N of 4". Plain text, under 110 words. Judge only what the reply says.
Common mistakes when you ask AI for this
- Trusting the confident tone. The reply sounds helpful, so the AI gives a pass on policy. Try: Say: "The current window is 30 days. Check every claim against it."
- Missing the sneaky blame. The AI treats the last sentence as friendly. Try: Ask: "Quote the exact phrase for every fail."
- Writing long evidence. The AI paraphrases instead of quoting short evidence. Try: Say: "Evidence: a direct quote under 8 words."
- Free-form scoring. The AI writes paragraphs of feedback instead of a table and total. Try: Give the exact header and the Total line format.
- Getting markdown back. AI adds bold and headings by default, which show up as stray symbols where this is pasted. Try: Say "plain text, no bold, no headings, no asterisks".
What a strong prompt includes
- Asks for plain text
- Sets a length limit
- Gives the table header
- Asks for quotes
Related exercises
- Alert sales to a competitor's price change, facts only (Research & AI Search · Product managers)
- Announce planned maintenance in four labeled lines (Customer Support & Success AI · IT teams)
- Answer a late-checkout request using the tiered policy (Customer Support & Success AI · Customer support teams)
- Answer a price-increase objection without discounting (Customer Support & Success AI · Customer success managers)
- Approve or return expense claims against the policy (Productivity AI · Finance and accounting)