You did not lose the hour to the meeting. You lost it to the four minutes after the meeting, and the four minutes after the next thing, and the next. Death by a thousand small re-entries.

Most of us think we lose time the obvious way: a long call, a fat document, a commute. But the quiet, expensive leak is the switch itself. Every time you put one thing down and pick another up, you pay a toll. The toll is invisible, it never shows on a calendar, and it compounds. This piece is about why that toll exists, roughly how big it is, and how you can put a number on your own version of it without buying anything or installing a single guilt-inducing tracker.

The brain does not have a clutch

Your attention is not a gearbox; it is a slow tide. When you stop working on Task A and start Task B, part of your mind stays behind. The psychologist Sophie Leroy named this effect "attention residue": a residual film of Task A that keeps clinging to your thoughts while you are supposed to be doing Task B. You are physically present at the new task, eyes on the new screen, but a slice of your working memory is still litigating the old one.

This matters because attention is the actual bottleneck of knowledge work. You can have all the time in the world and still produce nothing if your attention arrives in shredded little pieces. A surgeon does not operate for ninety seconds, answer an email, then return to the incision. We accept that some work demands continuity. The trick is noticing that your work demands it too, even when it looks like it could survive interruption.

The cost has three parts, and only one of them is obvious. First there is the reorientation: where was I, what was the next line, which tab held the thing. Second there is the residue: the lingering pull of the abandoned task, which quietly degrades the quality of what you do next. Third there is the error tax: switches are where mistakes are born, because you reload the wrong context or forget the half-finished step. The visible part, the reorientation, is the smallest of the three. The residue and the errors are where the real money goes.

Batching beats juggling, and it is not close

Doing similar things together is cheaper than alternating between different things. This is not a productivity-guru invention; it is just how setup costs work. Taiichi Ohno and the Toyota Production System obsessed over changeover time: every time a machine switched from making part A to part B, you paid for the re-tooling. Reduce the number of changeovers and you reclaim enormous capacity without working anyone harder. Your brain has changeover costs too. They are just harder to see because nobody hears the gears grind.

Imagine you have ten small writing tasks, each needing twenty minutes of actual work. If you do them back to back in one block, you pay the entry cost once: you warm up, you find your voice, and you ride the momentum. Now imagine you interleave them with ten unrelated admin tasks, switching after each one. Suddenly you pay an entry cost twenty times instead of twice, and each entry is polluted by residue from whatever you just left. The total clock time balloons even though the "real work" is identical.

The numbers above are made up on purpose, to show the shape, not to quote a study. The honest claim is narrower and sturdier: every extra changeover adds a fixed-ish penalty, so total time scales with the number of switches, not just the amount of work. If a single switch costs you, say, a few minutes of reorientation plus a quality dip you barely notice, then forty switches a day is most of an hour gone before lunch and a fog over everything you touched.

The decay curve nobody shows you

Residue does not vanish the instant you switch; it fades. Picture your effective focus on the new task right after a switch. It starts low, because part of you is still elsewhere, then climbs back toward full as the old task loosens its grip. The recovery is not instant and it is not linear. The first stretch after a switch is the most contaminated, which is exactly the stretch most people fill with "quick" reactive work before settling in.

The cruel part is that frequent switching keeps you permanently in the shallow end of that curve. If your recovery takes a while and you switch before you reach the top, you spend your entire day operating at partial focus and never realize it, because partial focus feels totally normal from the inside. You are not lazy and you are not slow. You are just never allowed to finish climbing.

This is also why "I multitask fine" is usually a self-report you cannot trust. Kahneman and Tversky taught us that intuition about our own mental performance is unreliable, especially when the cost is hidden and the activity feels busy. Switching feels productive precisely because it generates motion. Motion is not output.

Measuring your own switching tax

You cannot manage what you refuse to count, so count it crudely. You do not need a fancy app. You need one number: how many times per day you switch primary task, and roughly what each switch costs you. Here is a lightweight protocol you can run for one ordinary week.

• Tally the switches. Keep a notepad or a single text file open. Every time you change your primary task (not glancing, but actually starting something else), make a tick. Include the sneaky ones: opening email "for a second," checking a message, answering a quick question. At day's end, count the ticks.
• Time three honest re-entries. Three times a day, when you return to deep work after a switch, start a timer and note how long until you genuinely feel back in it. Not "I have the file open" but "I am thinking clearly about this again." Average those numbers.
• Multiply. Switches per day times your average re-entry time gives a conservative floor for your daily switching tax. It is a floor because it ignores the quality dip and the errors, which you cannot easily measure but absolutely exist.

Then do the arithmetic that will sober you up. Suppose you tally thirty-five switches in a day and your honest re-entry averages three minutes. That is over an hour and a half of pure reorientation, every single day, before counting a single mistake caused by reloading the wrong context. Across a working year that is weeks. Frame the numbers as your own measured estimate, not gospel, but even halved they are alarming.

What to actually do about it

Reduce the number of changeovers before you try anything clever. Most advice jumps straight to focus hacks, but the highest-leverage move is simply switching less often. Batch your shallow work: handle messages in two or three planned passes a day instead of a trickle of forty. Group errands of the same type. Cluster your meetings so the gaps between them are large enough to actually use. This is Ohno's changeover logic applied to your skull.

Protect a daily block where switching is forbidden by you, for you. A timebox is the practical tool here. Pick a stretch, ninety minutes is a good target, and treat it like the surgeon treats the incision: no re-entries allowed. Tiago Forte and others talk about this as creating an environment where deep work is the path of least resistance rather than an act of heroism. The point is not to be hardcore. The point is to stop paying the toll over and over.

Make your switches visible so they stop being free. The single act of tallying changes behavior, because once a switch costs a tick mark and you can see the total climbing, your brain starts asking "is this worth a tick?" That question alone kills a surprising number of pointless switches. You are turning an invisible cost into a visible one, which is the whole quantified-self game in miniature.

A few switches are unavoidable and even healthy; the goal is never zero. The goal is to stop bleeding hours you never agreed to spend.

The takeaway

Context switching is the most expensive thing you do that never appears on any invoice. It hides inside the gaps, it feels like work, and it quietly drags every task down a notch. The good news is that it is measurable. Spend one ordinary week counting your switches and timing your re-entries, multiply the two, and stare at the number. It will not be precise, and it does not need to be. It just needs to be big enough to make you rethink the next time you reach to "just quickly" check something. Protect a block, batch the rest, and let the residue finally burn off. Your best thinking is waiting at the top of a curve you have probably never reached.