Imagine a Monday Slack post: Agent B is at 78% CSAT, Agent A at 94%. Someone asks whether B needs coaching.
Then you look at the queues. B owns enterprise billing and refunds—conversations that already carry most of the team's DSAT. A handles password resets and plan changes. Same scorecard, different jobs. A team report that only ranks agents would have told you the wrong story. An agent report that ignored mix would have done the same.
Supportman sends both team-level and per-agent reports to Slack, so you can zoom between the system and the individual without building anything.
Team reports diagnose systems, demand, and shared outcomes. Agent reports help one person see their patterns and pick a development priority. Collapse both into one public ranking and you get a number that is neither fair nor actionable.
What the team report should answer
Leaders should finish a team report able to name one operational move: restaff a queue, fix a handoff, escalate a product pattern, or change a policy. If the report cannot support that kind of decision, it is a dashboard dump.
What demand reached us?
- Total conversations and day-by-day volume
- Volume by channel or queue
- Issue mix shifts versus the prior week
How did the system respond?
- First-response and time-to-close distributions (median plus a tail, such as p90)
- Backlog and age of oldest open conversations
- Handoffs and reopen or repeat-contact indicators
What did customers report?
- Rating count and response rate (not only the average)
- CSAT distribution
- DSAT themes with example conversations
What did quality evaluation find?
- Overall quality and attribute trends
- Issue types tied to weak scores
- Systemic product, policy, or workflow factors
The unit of action is the team or organization. Prefer shared interventions over naming individuals in the public view.
What the individual report should answer
An agent report should answer four questions in about five minutes. It is a development input, not a performance tribunal.
What work did I handle?
Show volume and mix—queue, issue type, plan tier—so scores sit next to context. Without mix, a hard queue looks like weak skill.
What patterns appear in my conversations?
- Recurring strengths worth repeating
- Rubric-attribute trends over the period
- Customer comments, positive and critical
- Two or three representative examples, not a dump of every ticket
What should I practice next?
One specific behaviour for the next week or two. “Be more empathetic” is not a goal. “On billing disputes, restate the disputed charge before citing policy” is.
What contribution should be recognized?
Include customer praise, strong handoffs, knowledge sharing, and quiet work that never shows up in CSAT—macros fixed, macros retired, coverage for a teammate's queue.
The unit of action is individual development.
Why context matters
Two agents can diverge on CSAT or resolution time because they handle different queues, languages, customer plans, issue complexity, escalation paths, or working hours. Raw ranking assumes comparable work. Often it is not.
Before treating a gap as a coaching problem:
- Require a usable sample. Treat fewer than about 15–20 rated conversations in the period as directional, not diagnostic. Tiny samples swing on one angry customer.
- Compare within similar mix—same queue or issue family—before comparing across the whole team.
- Separate customer ratings from QA evaluation. They answer different questions and often disagree.
- Read the distribution, not only the average. One 1-star ticket can flatten a week's mean.
- Check stability across two or three weeks. A single-week dip is usually noise or a demand spike.
If Agent B's CSAT is lower but their QA on comparable conversations matches peers, coach the queue design or product friction—not the person.
Use top-agent recognition carefully
Recognition can reinforce good work. A permanent CSAT leaderboard often rewards easy queues and punishes people who take the hard ones.
If the team report includes a callout:
- Rotate the dimension each week—resolution quality, communication improvement, a standout customer remark, escalation ownership, knowledge contribution.
- Credit teammates when the win depended on a handoff or shared playbook.
- Require a minimum sample (same 15–20 rated conversations or reviewed tickets bar) before crowning anyone.
- Never treat one week as a final judgment.
- Name the behaviour, not just the score: “Cleared the Friday billing spike without dropping first-response quality,” not “#1 CSAT.”
The goal is to show what good looks like under real constraints—not to run a weekly contest.
Connect the agent report to one goal
A field study of sales and customer-service representatives found that feedback seeking and self-efficacy related to work performance through feedback-based goals. Read the feedback-seeking field study.
The practical lesson: do not hand someone a scorecard and hope. Connect evidence to one behaviour they can practise. For the fuller coaching loop—gap, example, practice, follow-up—see the support QA coaching plan.
Example:
Evidence: Four reviewed billing disputes opened with a policy paragraph before confirming which charge the customer was questioning.
Goal: On the next 10 billing-dispute conversations, restate the disputed charge in the first reply before citing policy.
Review: Pull those 10 conversations next Friday and check for the restatement—not for a CSAT bump yet.
One goal beats five rubric notes. Score targets (“raise CSAT two points”) skip the behaviour that produces the score.
Use team goals for shared constraints
When the same failure appears across many agents—broken handoff, missing macro, unclear ownership on refunds—individual coaching will not fix it. That is a team intervention.
Use the team report to choose one shared constraint for the week:
- New workflow or ownership rule
- Better documentation or macros
- Clearer routing
- Shared calibration on a rubric attribute
- Product or policy escalation with a named owner
Do not coach people around a system defect for months. If half the team fails the same step, change the step.
Keep developmental feedback private
Put the team report in a shared channel. Send the individual report to the agent and their manager—not to the whole team.
Private delivery creates room for context, questions, disagreement, and a development plan. Agents will not engage honestly with weak conversations if those tickets are about to appear in a public ranking.
Aggregate themes can still return to the team: “Billing disputes are weak on confirming the charge before policy” is useful. Naming who missed it is not.
A simple routing rule
Run both reports on the same cadence—usually weekly—so the system view and the person view describe the same period.
- Team channel: volume, responsiveness, ratings with denominators, quality themes, one recognition callout, one operational action.
- Agent + manager DM: personal mix, patterns, one practice goal, personal recognition.
- Manager review: for anyone flagged, check sample size and mix before scheduling coaching. If the pattern is team-wide, skip the 1:1 and fix the system.
Leaders stay focused on operations. Agents get feedback they can use without performing for a leaderboard.
Quick comparison
| Team report | Agent report |
|---|---|
| Shared channel | Private to agent and manager |
| Diagnoses systems and demand | Supports individual development |
| Tracks team outcomes | Shows personal patterns and mix |
| Assigns one operational action | Creates one coaching goal |
| Recognizes team-visible wins | Recognizes personal contribution |