Coaching

Coaching is the recurring, usually one-on-one conversation where a team lead reviews real interactions with an agent, reinforces what worked, and agrees one specific behaviour to change before the next session. It is the step that turns QA scores, CSAT verbatims, and call recordings into actual behaviour change — without it, all that measurement is just scorekeeping.

It is a practice, not a metric, though teams track cadence and completion. Distinguish it from QA, which measures the work, and calibration, which aligns the reviewers: coaching is what you do with the measurement. The effective session is specific — tied to a named conversation, not a vague trend — timely, honest about both strengths and gaps, and ends with a single next step the agent owns.

What it hides: coaching is the first thing cancelled when the queue spikes, so "we coach every week" on the org chart is often "twice last quarter" in practice, and a logged session count says nothing about whether anything useful was said. A meeting that just reads scores back to an agent is a status update wearing coaching's name; the feedback that actually changes behaviour is far rarer than the calendar invite implies.