How to measure VA performance with three metrics (not a timesheet)
Measure quality (errors or rework per output), speed (cycle time from trigger to done) and consistency (percentage of cycles delivered on the agreed rhythm). Define each per task, not per person, and review weekly for the first month then monthly. Avoid screenshot and keystroke tracking: it measures presence, damages trust, and tells you nothing about whether the outcome arrived.
The three metrics, made concrete
- Quality — "fewer than one correction per ten replies", "zero pricing errors"
- Speed — "inbox cleared by 11:00 local", "report delivered Monday 09:00"
- Consistency — "48 of 52 weekly cycles delivered on time"
The weekly report that replaces check-ins
Ask for five lines every week: what was completed, what is in progress, what was escalated, what took longer than expected, and one suggested improvement. That last line is the difference between an assistant and a system owner.
What not to measure
Hours logged, screenshots, activity scores. They optimise for looking busy. If the outcome arrives on time and to standard, how long it took is the assistant's problem to optimise — and their incentive to build a better process.
When numbers say the process is wrong
If quality drops on one task only, the SOP is unclear. If speed slips across everything, scope has grown past capacity. If consistency slips at the same point every week, the trigger is competing with something else in the calendar.
Common follow-up questions
How often should I review?
Weekly for the first month, then monthly, with a quarterly scope review.
What if performance drops after a good start?
Usually scope creep. Re-list every task they now own and compare it to what was agreed at week four.