Agent Performance Dashboard Synthesizer

What it does

It takes a call center team's (or a single agent's) raw KPIs — AHT, FCR, CSAT, Service Level, occupancy, adherence, shrinkage — and converts them into an interpreted decision: "who should fix what, in which order?" The value is not echoing the table back; it is binding each number to a root cause and the next coaching move.

The core mechanic is a root-cause split: it uses decision rules to separate whether the problem is capacity-, behavior-, or measurement-driven. If occupancy is ≥88% and SL is below target, it is an FTE/roster (capacity) problem — coaching the agent is a misdiagnosis. Normal occupancy but adherence <85% is behavior/discipline. High FCR with low CSAT is attitude/tone; both low is a process/authority problem. It reads each metric as "number + direction + meaning" ("AHT 7:20, 22% over target, eating capacity"), produces a weighted Operations/SL Score (0-100), and uses Pareto to pick the single biggest lever.

When to use it

• When the KPIs in hand can't answer "good or bad, why, what to do."
• Before a performance review, coaching session, or shift retrospective that needs a one-page synthesis.
• When you need to isolate the root cause of an SLA drop (capacity / behavior / measurement).
• When a naive read would say "cut AHT, coach the agent" but the real issue is structural.

Method / frameworks

• Erlang-C (queuing theory): derives expected SL/occupancy from volume×AHT×agents; the observed-vs-expected gap anchors the capacity/behavior split.
• COPC CX Standard: audits whether occupancy sits in the sustainable band (75-85%), shrinkage accounting, and QA-metric consistency; >85% occupancy is flagged as burnout risk, not efficiency.
• SQM FCR operating philosophy: treats FCR as the primary directional metric — each +1% FCR ≈ +1% CSAT and ~1% cost reduction; low CSAT is diagnosed through FCR first.
• Schedule Adherence + Shrinkage accounting: reads adherence at the interval level (day-average hides peak collisions); no FTE recommendation without accounting for 30-35% shrinkage.
• GROW coaching model: turns each critical finding into goal + current reality + one concrete option; behavior-named, never agent-named.
• Pareto (80/20): picks the single lever instead of chasing all metrics at once; the rest go on a watch list.

Output is in plain language and never leaks internal identity or agent names. Best-effort degradation: if a source fails, the skill continues and writes the gap explicitly into gaps — no silent skipping.