Study finds Copilot users spend about two fewer hours a week on email
Reviewed by Helen Jones
A field experiment across 66 firms randomly gave Microsoft 365 Copilot access to 7,137 knowledge workers for six months. In the second half of the trial, the 80% of treated workers who used the tool spent about two fewer hours a week on Outlook email, a roughly 17% cut, and also reduced time working outside regular hours. Looking at everyone offered access, not only active users, email time still fell by about 1.4 hours a week, or 12%. Researchers did not find clear changes in how many tasks people did or what kinds of work they took on beyond those time savings. The paper, updated in 2026, measures integrated Copilot in email, meetings, and writing apps rather than vendor case studies alone.
Before this kind of evidence, many rollouts treated Copilot as a general productivity layer and judged success by licence counts, prompt volume, or stories of polished drafts. Teams often tried the same vague prompts across Word, Excel, Teams, and Outlook, then blamed the tool when results felt uneven. What changed is a clearer, narrower signal: the measurable win showed up as less time in email and less after-hours work, not as a rewrite of the whole job. That pushes adoption away from hype about task recomposition and toward honest playbooks that treat Outlook triage, thread summaries, and draft checks as the first repeatable use cases, with human review still required before anything is sent.
Analysis
Treat this as evidence to test, not proof that Copilot already pays for itself everywhere. For the next two weeks, pick one team email workflow (thread summary, reply draft, or inbox triage), record baseline Outlook time and rewrite effort, then run the same workflow with Copilot plus a short review checklist and compare hours and quality of what actually gets sent.
Source note
Pulse published by Collab365 Spaces, reviewed by Helen Jones on . Cite as "Study finds Copilot users spend about two fewer hours a week on email", Collab365 Spaces. 2 sources referenced.