Article
Measuring agent performance without breaking morale
Performance measurement in support goes wrong when it tracks output that agents cannot fully control. A better system measures what agents are actually responsible for.
March 5, 20262 min readBy Hashir Imran
Most support performance dashboards measure speed and volume because those are the easiest things to count. The trouble is that those numbers are heavily influenced by queue routing, channel mix, and shift timing, none of which the individual agent controls. Reward the wrong inputs and the team will optimize for them anyway, often at the cost of quality.
Three things worth measuring that are actually fair
- First-contact resolution within categories the agent owns. Comparing across categories is unfair because complexity varies, but within a category it is meaningful.
- Reopen rate. A conversation that closes and then reopens within a short window is usually a real signal about quality, not speed.
- Internal consistency. How often does the same agent apply the same policy the same way across similar cases? This is more important than raw output for long-term trust.
What to stop measuring on its own
- Average handle time, in isolation. It rewards rushing and punishes the agents who patiently solve hard cases.
- Total tickets closed per day, in isolation. It rewards channel mix more than actual performance.
- CSAT in tiny samples. It is noisy, biased, and easily skewed by issues that have nothing to do with the agent.
A useful frame
Good performance metrics make the right behaviour the easy behaviour. If your top performers feel punished by the metric, the metric is wrong, not them.