Measurement guide
How to measure time saved by Microsoft 365 Copilot
Start with one task and count the time people spend checking the answer. Yes, checking counts.
“Copilot saves 30 minutes a day” sounds very precise. It often comes from people guessing across a whole day or week. That can show a pattern, but memory is messy and workdays are not identical.
You will get a better answer by measuring the same kind of task several times. Try meeting summaries, client briefings, or project updates.
Use three levels of evidence
Quick estimate
Ask people how long the task usually took before and how long it takes now. Use this to find promising tasks. Label the result as self-reported.
Task sample
Record 10 to 30 comparable tasks before and after. Include prompting, editing, checking, correcting, and waiting. Report the median and range.
Operational result
Check whether the team completed more useful work, reduced a queue, worked less overtime, or improved quality. Saved minutes alone do not prove any of these results.
Count the full task
If the first draft is 15 minutes faster but checking adds 9 minutes, you saved 6 minutes. Report 6. The bigger number looks nicer on a slide, but it is still wrong.
Do not turn saved hours into pretend money
Suppose 100 employees say they save 20 minutes on three days each week. That adds up to 100 hours. It does not mean 100 paid hours suddenly disappeared from the budget.
The time may come in useless little pieces. People might spend it on better preparation, a shorter queue, learning, or less overtime. Decide what the time is meant for, then check if that happened.
Participants reported a median saving of 12 minutes on the weekly project update. Quality scores were unchanged. Most used the time for checking project risks. We have not found a cash saving.
Ask these five questions
- Which exact task became faster?
- How comparable were the before and after examples?
- Did you include checking and failed attempts?
- Who saved little or no time?
- What happened to the available time?