The four support metrics worth reporting to your leadership team

Support dashboards tend to report everything and explain nothing. If you are automating part of your support workflow, four numbers carry most of the signal — provided you also know how each one can flatter you.

Deflection rate is the share of incoming conversations resolved without a human. It is the headline number and the one most often overstated. A conversation where the customer gave up and left is not deflected, but a naive counter records it identically to a genuine resolution. Only count deflection where the customer either confirmed the answer helped or did not return with the same question within a few days.

Resolution rate measures how often the agent produced a complete answer, whether or not a human was also involved. It is a better measure of the agent’s quality than deflection, because it is not confounded by how many customers happen to want a human regardless. Track it per topic, not just in aggregate — an 80% average can easily hide a category where the agent fails almost every time.

CSAT on AI-handled conversations should be measured separately from your overall CSAT and compared against the human baseline for the same ticket types. Compared against your all-tickets average, automation will look artificially good, because it handles the easy questions. The honest comparison is like for like.

Knowledge coverage is the least reported and the most actionable: what share of incoming questions does your documentation actually address? Every gap is both an automation failure and a self-service failure, and unlike the other three metrics this one names its own fix. Ranked by volume, it is simply your content roadmap.

Reported together, these four tell a coherent story: how much work is being absorbed, how well, at what quality, and what to build next. Any one of them alone can be moved without improving the customer experience at all.