A rubric you own, applied the same way to every agent, with coaching evidence managers can use and a path from quality misses to the product themes behind them.
01 — INDEPENDENTGrade the change, not the vendor’s version of it
The AI agent vendor, helpdesk, or BPO that delivered your new support model isn’t a neutral grader of it. unitQ Support QA sits outside the delivery stack and applies your scorecard to every interaction, so quality is measured by the team that owns it rather than the team being measured.
02 — ONE SCORECARDScore AI agents, BPOs, and in-house teams the same way
Define the rubric once, covering tone, empathy, compliance, process adherence, handoff quality, and deflection accuracy, then apply it across every delivery mode and platform. Test it against real conversations before it goes live, so the first scores you publish are ones you can defend.
03 — COACHING EVIDENCECoaching evidence a manager can act on
Every scored interaction surfaces the moment that moved the score: the skipped verification, the escalation that never happened, the handoff that dropped the customer. Managers coach from specific evidence instead of a sample of tickets and a hunch, and bots face the same scrutiny as people.
04 — QUALITY → PRODUCTFrom a quality miss to the product theme behind it
Wired to unitQ Monitor, quality misses link to contact reasons and product themes, so a pattern in the support queue points at whatever in the product is causing it. That’s the evidence that lets you expand automation instead of capping it.