Prompts for Support QA, Coaching, and Knowledge Base

3 ready-to-use ChatGPT prompts for support qa, coaching, and knowledge base. Fill in the bracketed fields, paste into ChatGPT, and go.

Customer Support · 3 prompts

01

Review a Support Ticket Transcript for Coaching Opportunities

Review a support ticket transcript for specific coaching opportunities, distinguishing genuine skill gaps from one-off situational judgment calls.

You’ll needticket transcriptquality dimensions to assessagent experience levelknown team-wide coaching focus areas
Prompt
You are a support QA reviewer.

Review the transcript using the context below.

Context:
- Ticket transcript: [transcript]
- Quality dimensions to assess (accuracy, tone, resolution speed, empathy): [dimensions]
- Agent experience level: [experience]
- Known team-wide coaching focus areas: [focus areas]

Instructions:
1. Identify specific moments in the transcript tied to each quality dimension rather than a general impression score, so the coaching conversation can reference exact lines.
2. Distinguish a genuine skill gap (recurring pattern) from a one-off judgment call that was reasonable given the situation, since treating every imperfect moment as a gap creates unproductive nitpicking.
3. Note what the agent did well specifically, not just what to improve, since coaching that's all correction reads as purely critical and is less likely to land.
4. Calibrate expectations to the agent's experience level — a first-week agent and a two-year veteran shouldn't be held to identical coaching standards on judgment calls.
5. End with 1-2 specific, practice-able coaching points rather than a long list that's hard to act on.

Output format:
- Dimension-by-dimension findings with specific transcript references
- Genuine gaps vs reasonable judgment calls
- What the agent did well
- 1-2 specific coaching points

Missing information policy:
Ask up to 2 concise questions only if the transcript or quality dimensions are unclear.

Quality bar:
- References specific transcript moments, not general impressions.
- Distinguishes genuine gaps from reasonable judgment calls.
- Limited to 1-2 actionable coaching points.
02

Turn a Resolved Ticket into a Knowledge Base Article

Turn a resolved support ticket into a knowledge base article generalized for self-serve use, not written as a narrative of this one specific case.

You’ll needresolved ticket detailsissue frequency (how often this comes up)target audience for the articleknown variations of the issue
Prompt
You are a support content strategist writing a knowledge base article.

Write the article using the context below.

Context:
- Resolved ticket details: [ticket]
- Issue frequency: [frequency]
- Target audience for the article (technical, general): [audience]
- Known variations of the issue: [variations]

Instructions:
1. Generalize the article from this one customer's specific situation into a reusable format, since a KB article written as a narrative of one ticket doesn't read as self-serve guidance.
2. Write the title as a specific, searchable question or problem statement matching how a customer would actually search for this, not an internal-sounding label.
3. Cover the known variations of the issue so a customer whose situation differs slightly from the original ticket can still find their answer in the same article.
4. Match the technical depth to the target audience — don't assume technical fluency for a general-audience article.
5. Note where the article should link to a related article or escalation path if the self-serve steps don't resolve it.

Output format:
- Searchable article title
- Generalized problem description
- Step-by-step resolution
- Known variations covered
- Escalation path if self-serve doesn't resolve it

Missing information policy:
Ask up to 2 concise questions only if the ticket details or target audience are unclear.

Quality bar:
- Generalized, not a narrative of one ticket.
- Searchable title matching real search behavior.
- Covers known variations, not just the original case.
03

Design a Support Quality Scorecard That Reps Won't Game

Design a support quality scorecard resistant to being gamed on individual metrics, balancing speed, resolution quality, and customer sentiment.

You’ll needteam contextcurrent metrics trackedknown gaming patterns if anyquality prioritiesreporting cadence
Prompt
You are a support operations strategist designing a quality scorecard.

Design the scorecard using the context below.

Context:
- Team context: [team]
- Current metrics tracked: [metrics]
- Known gaming patterns if any (rushing to close tickets, avoiding hard cases): [gaming patterns]
- Quality priorities: [priorities]
- Reporting cadence: [cadence]

Instructions:
1. Balance speed metrics (first response time, resolution time) against quality metrics (resolution rate, CSAT, reopened-ticket rate) so optimizing purely for speed doesn't quietly hurt actual resolution quality.
2. Include a reopened-ticket or repeat-contact metric specifically, since fast-but-wrong resolutions often show up there rather than in first-response speed.
3. Flag any metric known to be gameable given the team's history, and note a paired metric that would catch the gaming.
4. Weight metrics differently for different ticket complexity, since penalizing an agent's average handle time for taking on the team's hardest tickets discourages picking them up.
5. Keep the scorecard simple enough that agents understand what it's actually measuring, not a black-box formula.

Output format:
- Balanced metric set (speed + quality)
- Reopened/repeat-contact metric included
- Gaming-resistant pairing notes
- Complexity-adjusted weighting approach

Missing information policy:
Ask up to 2 concise questions only if the current metrics or quality priorities are unclear.

Quality bar:
- Balances speed against quality, not speed alone.
- Includes a reopened/repeat-contact signal.
- Complexity-adjusted, not flat across all tickets.