You Don’t Have an Attention Problem. You Have a Context-Switching Problem.
Most people who feel like they “can’t focus anymore” don’t actually have a focus problem. They have an environment that interrupts them every few minutes and a set of habits that interrupt them the rest of the time, and no amount of willpower fixes a structural problem like that. Focus isn’t a personality trait you either have or lack. It’s a resource that gets spent, and context-switching is the single most expensive way to spend it.
Understanding exactly how expensive changes how seriously you take the problem — and once you see the actual numbers, “just try to concentrate harder” stops sounding like a plan.
The Number That Should Change How You Work
In one of the most cited studies in attention research, Gloria Mark and her colleagues at the University of California, Irvine, followed office workers throughout their actual workdays, tracking every interruption as it happened rather than relying on people’s self-reported memory of them afterward. The study found that after an interruption, it took workers an average of 23 minutes and 15 seconds to fully return to the original task, and that people didn’t return to it directly — they typically moved through roughly two other tasks first before getting back to what they’d originally been doing. Twenty-three minutes. Not the length of the interruption itself — the length of time it takes your brain to actually get back to where it was before the interruption happened.
Do that math against an ordinary workday. A handful of Slack pings, a couple of “quick questions,” an email that pulls you off-task for two minutes, and you’re not looking at a few minutes of lost time. You’re looking at the better part of your morning gone, structurally, before lunch — not because you’re undisciplined, but because that’s what the recovery cost of interruption actually looks like once someone bothered to measure it instead of guessing.
It gets worse before it gets better. Mark’s later research, following up on her original findings, found that roughly 44% of workplace interruptions are self-generated — not a colleague, not a notification, but the person’s own reflexive urge to check something else. That reframes the whole problem. You can’t fix context switching purely by managing other people’s demands on your time, because a large share of the damage is coming from your own hand reaching for your phone.
Why “Just Concentrate Harder” Was Never Going to Work
The American Psychological Association’s research on task-switching found that shifting attention between tasks can cost as much as 40% of a person’s productive time, because the brain doesn’t run two complex tasks in genuine parallel — it toggles rapidly between them, and each toggle carries a real cost in both speed and accuracy that compounds the more often it happens. Combine that with Mark’s 23-minute recovery finding and you get a fuller picture of why the feeling of being constantly busy but never quite getting anywhere isn’t an illusion. It’s the accumulated cost of switching, measured and named, not a personal failing to feel guilty about.
Step One: Measure Your Actual Switches for One Day
You can’t fix what you haven’t actually seen, and most people wildly underestimate how often they switch. For one day, keep a simple tally — a tick mark on a sticky note every time you move from one task or app to another, whether someone else pulled you away or you did it yourself. Don’t change anything yet. Just count.
Most people who do this exercise are startled by the number, and that number is the actual diagnostic that tells you where to focus your fix. If most of your switches are self-generated, the solution lives in your own habits. If most are external, the solution lives in how you manage other people’s access to your attention.
Step Two: Batch Your Inputs Into Windows, Not a Continuous Stream
Email, chat messages, and notifications don’t need to be processed the instant they arrive, even though every app is designed to make it feel that way. Pick two or three specific windows during the day — say, mid-morning, after lunch, and late afternoon — and check messages only during those windows. Outside of them, the notification badge can wait; it isn’t actually going anywhere.
This single change does more to reduce self-generated interruptions than almost anything else, because it converts a continuous, reflexive stream of small pulls on your attention into a handful of deliberate, contained sessions. You’re still responsive. You’re just responsive on a schedule you set, rather than one dictated by whichever app pinged last.
Step Three: Build a Transition Buffer Between Tasks
Part of what makes the 23-minute recovery cost so brutal is that most people move directly from one task into the next with zero buffer, carrying the mental residue of the last thing straight into the next one. A short, deliberate transition — even just two or three minutes to jot down where you left off, close unrelated tabs, and take a breath before opening the next thing — measurably reduces how much of the previous task’s residue follows you into the new one.
This feels like wasted time in the moment. It isn’t. It’s the cheapest insurance available against dragging an unfinished mental thread into work that needs your full attention to do well.
Step Four: Cut Off Your Own Self-Generated Interruptions First
Given that close to half of interruptions come from the person doing the work rather than their environment, this step deserves more attention than most advice gives it. Move your phone out of arm’s reach during focus blocks — not just on silent, actually out of reach, since the mere visibility of a phone has been shown to pull on attention even when it doesn’t buzz. Close browser tabs unrelated to the task in front of you before you start, rather than trusting yourself not to click over. Treat the urge to “just quickly check” as exactly what it is: a habit loop, not a genuine need, and one that gets weaker every time you don’t act on it.
Step Five: Redesign the Workload, Not Just Your Willpower
Individual habits only go so far when the underlying workload itself is structured to demand constant switching. The CDC’s National Institute for Occupational Safety and Health identifies hectic, fragmented work with little sense of control as a defining source of job stress, and recommends job redesign — rather than individual coping strategies alone — as a primary prevention approach. That’s a meaningful distinction: NIOSH’s own guidance treats fixing the structure of the work as more effective than simply asking individuals to cope better with a workload that was never reasonable to begin with.
Practically, this means advocating for real changes where you have any influence over them — fewer standing meetings that fragment the day into unusable slivers, clearer boundaries around who can interrupt whom and when, and workloads sized to what a person can actually do in focused, uninterrupted stretches rather than squeezed in around constant demands.
Where AI Tools Actually Fit Into This
AI tools can help absorb some of the switching cost, but only in specific, limited ways. Having an assistant triage and summarize a backlog of messages during your batched checking windows, rather than you reading each one as it arrives, keeps you out of the continuous stream without losing track of what’s in it. We’ve written more broadly about the tasks worth keeping firmly in human hands rather than delegating to AI, and deciding what actually deserves to interrupt you is squarely one of them — a model can help you sort the incoming noise, but the judgment about what’s genuinely urgent still belongs to you.
It’s also worth being honest that AI tools can add to the switching problem as easily as they solve it. Our piece on the cognitive cost AI tools quietly add to a knowledge worker’s day covers exactly this — a chatbot tab open alongside everything else is one more place attention can leak to, not automatically a fix. If you’re building the habits above into a broader system, our guide to building a personal productivity system that actually survives a bad week pairs directly with this one; context switching is precisely the failure mode a good weekly structure and protected focus blocks are designed to prevent. And if part of your work involves running AI-assisted meetings, our guide to using AI for meeting notes responsibly touches on a related discipline — knowing which conversations deserve full, undivided attention rather than half-attention with a bot quietly taking notes in the background.
A Quick Check for Whether It’s Actually Working
Give the changes above a week, then look honestly at what changed rather than how motivated you still feel. Did your self-tally of switches actually drop, or did you just start feeling more guilty about the same number of switches? Are your batched checking windows still holding, or did the third day of a stressful week quietly dissolve them back into constant checking? The goal isn’t zero interruptions — that’s not realistic for almost anyone’s job. It’s a meaningfully lower number, with real, protected stretches where the 23-minute recovery cost simply doesn’t get triggered in the first place, because nothing pulled you away to begin with.
Frequently Asked Question
How long does it actually take to recover focus after an interruption?
Research from Gloria Mark at UC Irvine found that after an interruption, it takes an average of 23 minutes and 15 seconds to fully return to the original task, and people typically move through about two other tasks before getting back to what they were originally doing.
Are most interruptions caused by other people or by ourselves?
A significant share come from the individual, not their environment. Follow-up research from Gloria Mark found that roughly 44% of workplace interruptions are self-generated, meaning managing external demands alone won’t solve the full problem — personal habits around checking phones and switching tabs matter just as much.
Why is multitasking so much less efficient than it feels?
The American Psychological Association’s research found that switching attention between tasks can cost as much as 40% of a person’s productive time, because the brain toggles rapidly between tasks rather than genuinely processing them at the same time, and each toggle adds measurable cost to both speed and accuracy.
What’s the most effective first step to reduce context switching?
Measuring how often you actually switch for one day, before changing anything. This shows whether most of your interruptions are self-generated or come from external sources, which determines whether the fix should focus on personal habits or on managing how others access your time.
Does batching email and messages actually help with focus?
Yes. Checking messages during a few set windows rather than continuously converts a constant stream of small interruptions into contained, deliberate sessions, which meaningfully reduces the number of times the roughly 23-minute recovery cost gets triggered throughout the day.
Is context switching just a personal willpower problem?
Not entirely. The CDC’s National Institute for Occupational Safety and Health identifies hectic, fragmented workloads as a structural source of job stress and recommends redesigning the work itself, not just individual coping strategies, as the more effective long-term fix. Personal habits help, but workload structure matters just as much.
The Conclusion
Context switching isn’t a discipline problem you can will your way out of — it’s a measurable cost, roughly 23 minutes of recovery per interruption, compounding across a day that likely has far more interruptions in it than you’ve ever actually counted. Nearly half of those interruptions are self-generated, which means the fix starts with your own habits, not just other people’s demands on your time.
Batching inputs, building real transition buffers, and pushing for workload structures that don’t demand constant switching in the first place will get you further than any amount of trying to concentrate harder on a day that was never structured to allow it. Measure it, adjust it, and check back in a week — the data will tell you whether it’s actually working.
