Agent coaching packs

When to use it

Reach for this when someone says any of these — they are the phrases the skill itself triggers on:

• “prepare a coaching session for X”
• “what areas does X need to improve”
• “areas of markdown for this agent”
• “what coaching opportunities stand out”
• “build a coaching agenda from these tickets”

How it works

The method, in the order the skill runs it. The full procedure — tables, worked examples and the edge cases — is in the skill itself.

1. Step 1: establish what the data can support

   Do this before reading a single evaluation, because it determines what kind of pack you can honestly write.

2. Step 2: check the evidence is safe to coach from

   Exclude, and say you excluded: Superseded evaluations. Re-evaluated tickets leave old records behind, Tickets the agent did not meaningfully handle. Reassignments, one-line, Anything caused by something outside their control.

3. Step 3: find themes, not criteria

   Criterion names are rubric artefacts. "Probing & Understanding — partial" is not something anyone can act on.

4. Step 4: include the counter-evidence

   Two things belong in every pack and are almost always missing: Genuine strengths, from the data. Not a compliment sandwich, A high-scoring example of the same behaviour you are coaching. If they.

5. Template

   Confidence: score <x> ± <y> points at n=<n>. <What this supports and what it doesn't.> Mix note: <how their work mix compares to the comparison group, or "not compared">

Related skills

• Agent tooling frictionFind the tools, permissions and dead ends slowing agents down
• Attrition early warningSpot agents at risk of leaving, without surveillance
• Career pathsBuild progression that isn’t just leaving the phones
• What to hire forHire for what actually predicts good support work
• Bonuses and incentivesDesign bonuses that don’t wreck the metric they pay on
• New hire ramp-upSee how long new agents take to get good, and where they stall