The 23-Minute Tax You Pay Every Time You Switch

In 2008, Gloria Mark at UC Irvine published a number that people have been citing and misquoting ever since.

Twenty-three minutes and fifteen seconds.

That's the average time it takes to fully return to a task after being interrupted. Not the interruption itself — that might be thirty seconds. The recovery. The process of locating where you were, what decision you were in the middle of, what the problem actually was before something pulled you out of it.

That was 2008. Most of what's in your current stack didn't exist yet.

The word "interruption" is already outdated

The original research defined interruptions as external events: a phone call, a colleague stopping by, a notification that pulls you away. You're working on A, something forces you into B, and then you have to find A again.

That's not the only version anymore.

Opening another tool is an interruption too. Closing a document and switching to ChatGPT. Leaving a meeting and jumping to Slack. Tabbing from an email draft to Perplexity to check one thing, then back. Each of these is a context switch — your brain leaving one problem, entering another, and then trying to recover the first one. The recovery cost doesn't care whether you were interrupted by someone else or by yourself.

A 2005 paper from the same research group found that knowledge workers switched working contexts every ten and a half minutes on average, passing through 2.3 intermediate tasks before returning to the original one. You go to "check one number" and come back two detours later, slightly less sure of what you were writing.

That's not a willpower issue. That's how switching works.

And that was before most people had five AI tools open at the same time.

AI tools didn't reduce the switching. They added a new decision on top of it.

Every task now comes with a pre-task: which tool?

ChatGPT or Claude? Search or Perplexity? Rewrite it yourself or ask the model? None of these are hard decisions — each one takes maybe three seconds. But three seconds happens dozens of times a day, before the work begins, before you've settled into the problem. And none of it shows up in any time log, because nobody writes "spent ten minutes deciding which app to open" in a daily summary.

But those ten minutes are real.

More than that: every time you make that decision, you're pulling yourself out of the last thing you were doing. More tools means more pull-outs. The tools themselves aren't the problem — each one is probably doing what it's supposed to do. But together they create a structure where you spend the whole day taking off and landing, taking off and landing.

Mark's later research tracked how long knowledge workers stayed on a single screen before switching. In 2004, the average was two and a half minutes. By 2012, it had dropped to seventy-five seconds.

That was thirteen years ago.

Why afternoons are harder than they should be

There's a phenomenon a lot of people have experienced but nobody's explained clearly.

Your best thinking usually happens in the morning. Not because you're trying harder, not because something specific derails the afternoon. It's because context-switching costs are cumulative. Each switch doesn't exhaust you immediately — but they stack, and by two or three in the afternoon they've accumulated into something that feels like slowness. Not tiredness exactly. More like the gears are turning but not catching.

The research has a name for this: resumption lag. The idea is that even after you return to a task, you haven't fully returned. You're working from a slightly reconstructed version of where you left off, and the gap is small enough that you don't notice it — until the afternoon, when you've accumulated enough of them that the deficit becomes visible.

This isn't an excuse. It's a mechanism.

The research doesn't offer a solution. But it points somewhere.

Mark's own conclusion was that interruption management is a design problem, not a behavioral one.

Which means: if your tool structure requires constant switching, telling yourself to focus harder won't fix it. You're not fighting your attention — you're fighting a system designed to require context switches. The system has the home advantage.

What you can reduce are the transitions that don't belong to the work itself.

Setup decisions. Physical friction. The time before the day starts — finding the cable, locating an outlet, adjusting the screen, deciding if this spot is workable. Each of these is small. But they happen before you've started, and they use the freshest part of your cognitive resources to produce nothing.

You start working already twenty minutes in, with nothing to show for it.

That's a smaller problem than the full architecture of modern knowledge work. But it's also the one that's actually solvable.

Part of Mukiya's World Safety & Health Day 2026 series. Code WORKSAFE — 15% off until May 1.mukiya.com/work-safe-smart

Sources: Mark, G., Gudith, D., & Klocke, U. (2008). The cost of interrupted work: More speed and stress. CHI '08. / Mark, G., Gonzalez, V. M., & Harris, J. (2005). No task left behind? CHI '05.