Agent performance
GET /performance adds up the daily figures for a day range (day_from, day_to) and
returns one row per group. Choose the groups with group_by (any of agent, team,
campaign, dialer, day) and narrow with employee_id, team_id, campaign_id or
dialer_id.
| You want | Request |
|---|---|
| Agent leaderboard | group_by=agent&sort=avg_score&min_scored=20 |
| One agent’s scorecard | employee_id=… |
| One agent’s daily trend | employee_id=…&group_by=day |
| Team comparison | group_by=team |
| Agents within each campaign | group_by=campaign&group_by=agent |
| Most fatal failures | group_by=agent&sort=fatal_rate |
| Most mis-coded calls | group_by=agent&sort=ai_disposition_mismatch_rate |
Each row has volumes (calls_total, scored, sales_count, vm_count), quality
(avg_score, fatal_rate, counts per band), behaviour (transfer_connect_rate,
avg_talk_seconds, ai_disposition_mismatch_rate) and review activity
(needs_review_count, override_count). avg_score covers full-length calls (60 seconds
of talk time or more); shorter calls are checked for compliance only.
team is the agent’s team on each day, so moving an agent does not rewrite history.
Rule by rule
Section titled “Rule by rule”GET /performance/rules shows how each QA rule went: pass rate for compliance rules,
average 1-5 score for KPI rules.
- An agent’s weakest rules:
employee_id=… - Who fails the recording disclosure most:
rule_key=C001&group_by=agent - How a rule trends:
rule_key=K003&group_by=day
The rule names behind each rule_key come from GET /directory/qa-rules.
Coaching
Section titled “Coaching”GET /coaching-cases lists coaching cases opened for agents, teams or rules, by QA staff or
by automatic detectors. Filter by status, employee_id or team_id, or sync with
updated_since.
Doing the maths yourself
Section titled “Doing the maths yourself”GET /facts/agent-daily returns the raw daily counts and sums per agent × campaign × dialer.
If you store them, always add up the parts before dividing:
- QA score:
Σ full_score_sum / Σ full_scored - Fatal rate:
Σ fatal_count / Σ scored - Band share:
Σ good_count / Σ scored(likewise fine, average, fair, poor) - Voice-machine rate:
Σ vm_count / Σ calls_total - Transfer connect rate:
Σ transfer_connected_all / Σ transfer_attempted_all
Never average daily or per-team averages. /performance does this for you.