Coaching conversations that don't feel like performance reviews
Slide a QA scorecard across the desk and the rep reads a verdict, not a lesson — here's how to turn the score into coaching that changes the next ticket instead of triggering a defence.
You slide the QA scorecard across the desk, and before you've said a word the rep's arms are folded. They're already reading it like a verdict — 84 out of 100, three categories glowing red — so every sentence you say next lands as an accusation they need to rebut. That isn't a coaching conversation. It's a performance review wearing a coaching lanyard, and both of you can feel the difference.
The score did that. A number on a rubric is a judgment by design — it ranks, it grades, it goes in a spreadsheet next to a name. Put it in front of someone and their nervous system reads how much trouble am I in, not what should I try on Tuesday. The information you actually want them to act on never makes it through the flinch.
A score tells you where you are. It never tells you what to do.
The deeper problem is that the number doesn't contain the coaching.
The industry-benchmark internal quality score sits around 88%, and most teams cluster within a few points of it. So telling a rep "you're at 84" locates them on a distribution and nothing more. It doesn't name a behaviour, doesn't point at a ticket, doesn't describe what a better reply would have looked like. The score is smoke. Coaching is about the fire, and you have to go into the actual conversation to find it.
A score tells a rep exactly where they landed and absolutely nothing about what to do next. Closing that gap is the entire job of coaching — and no number can do it for you.
Don't coach to the metric — especially not CSAT
The fastest way to turn a 1:1 into a tribunal is to open with a number the rep already knows they can't trust. Customer satisfaction is the usual culprit.
An agent genuinely performing at around 83% CSAT has roughly a one-in-six chance of landing at 72% or below in a given month, purely from small-sample noise. Pull that rep in to "work on their CSAT dip" and you're coaching a coin flip — and they know it. They know the survey volume is tiny; they know one furious customer with an unrelated billing complaint sank the week. Treat that number as a signal and you spend the only currency coaching runs on: credibility.
The mismatch runs deeper than noise.
Across more than 265,000 tickets, CSAT scores didn't correlate with QA scores — the two measure genuinely different things. Your QA rubric can't stand in for how the customer felt, and CSAT can't stand in for how well the rep worked the case. Coach each on its own terms, and never use one to indict the other across the desk.
Coach one behaviour, not the whole scorecard
The average support scorecard grades 14 categories; even the median is 8. Nobody can coach 14 things. A rep can carry exactly one behaviour from this conversation into their next shift — "acknowledge the problem before you explain the fix," or "confirm the account before you start troubleshooting." Pick the single change that would move the most tickets and leave the other thirteen boxes to the trend line. A conversation that hands someone one thing to try feels like help. One that walks all fourteen reds feels like a deposition.
Make the rep do the reviewing
Format is where "performance review" quietly becomes "coaching." Instead of reading someone their score, pull up a real conversation — one of theirs — and ask them to walk you through it. What were they trying to do here? What would they change? People are harder on their own work than you'd ever be to their face, and a fix a rep reaches themselves survives contact with Monday's queue in a way a fix you dictated never does.
Only about 44% of teams actually bring conversation-review feedback into their 1:1s. For a lot of reps, that means QA is a score that surfaces in a dashboard and never becomes a conversation at all — a grade with no teacher attached. Closing that gap is most of the job. The scorecard's value was never the number; it was the specific, reviewable moment the number was pointing at.
The tell
You'll know it worked by what the rep does next. A performance review ends with relief that it's over. A coaching conversation ends with them wanting to pull the next ticket and try the thing. Same score, same rubric, same manager — completely different room. The only variable is whether you spent the time defending a number or building the next reply together.
When two reviewers score the same ticket differently, the trust coaching depends on erodes before you've said a word — see calibration: why reviewers disagree. And if your rubric grades things that don't predict outcomes, coaching to it just teaches compliance; start with a QA rubric that predicts outcomes.