Does Copilot make you more efficient?

I’ve been reading Oliver Burkeman’s Four Thousand Weeks. It’s changed how I talk about Copilot with clients.

The usual pitch is simple: Copilot gives you time back. It drafts the email, summarizes the meeting, finds the file buried in SharePoint. Burkeman’s book raises a harder question. Does getting more efficient actually give you more time? Or does it just raise the bar for how much gets packed into the same week?

The efficiency trap applies to Copilot too

Burkeman calls this the efficiency trap. The idea: if we get efficient enough, we’ll finally clear the inbox and get ahead. In practice, the opposite happens. The faster email gets processed, the more email shows up. Technology increases throughput. It doesn’t reduce demand.

I’ve seen this with Copilot rollouts. A team cuts meeting prep from 30 minutes to five. A month later, the calendar is just as full. This time with more meetings, because now there’s room for them. That’s not a Copilot failure. Copilot did its job. The organization converted the time savings into more throughput, because that’s the default unless someone decides otherwise.

Here’s the distinction I’d draw for clients: Copilot doesn’t give you time back. It reduces the effort a task takes. Those aren’t the same thing. Confusing them is where a lot of the marketing oversells the tool.

Where it actually pays off

Burkeman isn’t arguing for doing less work overall. He’s arguing that time is limited, so we should be deliberate about what gets it. That’s a useful filter for Copilot projects.

Take meetings. Historically, everyone attends because attendance is how information gets distributed. With transcription and recap, that’s no longer true. The better question isn’t “can Copilot make this meeting more productive.” It’s “does everyone need to be here.” Cut a meeting from twelve people to three, and for the nine who skip it, that’s an hour back — not a 10% gain.

Search works the same way. Microsoft 365 Copilot is grounded through what Microsoft now calls Work IQ — a layer that connects Copilot and agents to email, files, chats, meetings, and people data the user already has access to. The interface shifts from digging through folders to just stating what you need.

One caution I give every client: Copilot inherits your information architecture. It doesn’t fix it. If your SharePoint is full of stale, redundant, or overshared content, Copilot surfaces that content faster: it doesn’t clean it up. Governance work such as permissions, sensitivity labels, retention; isn’t a separate project from your Copilot rollout. It’s part of the same one.

What it can’t decide for you

Agents raise the stakes further. An agent doesn’t just answer a question; it can retrieve information, reason over it, and take action across systems. That removes a lot of manual coordination between Outlook, Teams, SharePoint, and your CRM.

But no agent decides what’s worth doing. Copilot can summarize a hundred emails. It can’t tell you which relationships deserve your attention. It can draft ten versions of a deck. It can’t tell you whether the deck should exist. Those stay human calls.

My recommendation on measuring adoption

Most organizations track Copilot adoption through usage: active users, prompts, hours saved. Those numbers matter, but they only measure activity. I’d add a second set of metrics: meetings avoided, documents that didn’t need to get made, work eliminated rather than just sped up. That last one is harder to track. It’s also the one that tells you whether the rollout actually worked.

The goal isn’t to use Copilot to fit more work into the week. It’s to use it so people spend their time on what actually matters. That’s a narrower promise than most vendors make but it’s the one I’d stand behind.