Coaching and calibration
Two things that decide whether reviewing is worth doing at all: whether anything happens afterwards, and whether two reviewers would have scored the same call the same way.
Coaching
Coaching
A review ends with something the agent should do differently. Coaching is where that becomes a commitment with a date on it rather than a sentence in a form nobody returns to.
What a follow-up is
- The improvement area
- The one thing to work on. One, not five — an agent given five things to improve improves none of them.
- The recommendation
- What specifically to try instead. Concrete enough to practise.
- The date
- When somebody will look again. This is what separates coaching from a comment.
- The outcome
- Recorded when the follow-up comes round: better, unchanged, or no longer relevant.
Pick one thing
The temptation after a weak call is to list everything. Resist it. An agent who is asked to fix one habit this fortnight usually does; an agent handed a list treats it as a verdict on them rather than a task, and nothing changes.
Choose the thing that moves the outcome
If an agent greets badly and never agrees a next step, coach the next step. The greeting costs a point on the scorecard; the missing next step costs the deal. The owner’s Where quality is won and lost chart is the fastest way to see which criteria actually matter across your team.
The follow-up is the point
The coaching report tells an owner whether follow-ups are actually being closed. A team with good coverage, sensible scores and no completed follow-ups is measuring carefully and changing nothing — which is the most common way this kind of programme fails, and it is invisible unless somebody looks.
See Reports and exports.
What the agent sees of it
Their improvement area, the recommendation, and when it is due, on their own page. Not the private supervisor notes.
An agent should never be surprised by a follow-up in a performance conversation. If it is worth coaching, it is worth them being able to read it the week it was written.
Calibration
Several reviewers score the same call independently, then compare. It answers the question every quality programme eventually faces: does a 7 from one reviewer mean the same as a 7 from another?
If it does not, every figure built on those scores is noise — an agent’s average becomes a statement about which reviewer happened to get their calls.
How a session runs
- Somebody picks a call. An ordinary one is more useful than a dramatic one — the disagreements that matter are about everyday judgement.
- Each reviewer scores it separately, without seeing the others’ answers.
- The scores are compared, criterion by criterion.
- Where they disagree, the reviewers talk about why.
- The wording of the criterion gets fixed, or the shared understanding of it does.
Disagreement is the useful output
A calibration where everybody agreed tells you nothing you did not already assume. The criteria where reviewers split are the ones costing you data quality right now, and finding two of them is a successful session.
What disagreement usually means
| Pattern | Usually means | Fix |
|---|---|---|
| Everybody splits on one criterion | It is not scorable as written | Rewrite it as an audible behaviour |
| One reviewer is consistently higher | They are applying the scorecard more generously | Talk through the specific calls; it is usually one criterion, not a temperament |
| Disagreement about critical failures | The threshold was never agreed out loud | Decide it together and write it into the criterion |
| Wide disagreement on everything | The scorecard is too subjective to be useful | Cut it back to four audible criteria and rebuild |
How often
- Once in the first month, as soon as you have two reviewers — this is when the scorecard is loosest.
- Then monthly for a while, which is enough to keep reviewers aligned.
- Whenever a new reviewer joins, before their scores go into anybody’s average.
- Whenever the scorecard changes meaningfully.
- Whenever two reviewers’ average scores drift apart, which the queue page shows.
If you are the owner reading this
Calibration is where you find out whether your scorecard works. Running one yourself with your reviewers in the first month is worth more than any amount of rewriting criteria at a desk: you will discover two that nobody can score consistently, and fixing those two is the difference between figures you can act on and figures you cannot.
Then go back to Scorecards and criteria and rewrite them as behaviours.