What Research Actually Says About Email Interruptions

What research actually says about workplace interruptions tends to get lost in a relay race of misquotes. The most repeated figure — that it takes 23 minutes to fully recover from a single interruption — circulates on productivity decks and HR slide shows almost never with the original study cited and never with its sample size disclosed. The truth underneath is more useful than the shorthand: real research does show that interruptions carry measurable cognitive costs, that switching tasks imposes time penalties, and that the way email is typically used amplifies both. But specific numbers depend heavily on task type, study conditions, and how recovery is defined — and some of the claims you will encounter most often trace back to nothing checkable at all.

This is a careful read of what published research on workplace interruptions actually establishes. We look at where the 23-minute figure comes from and what it genuinely measures, what cognitive scientists have found about task-switching costs more broadly, what attention residue is and why it is a more durable finding than any single number, and which claims you should treat as folklore until someone produces a credible primary source. Every claim here names its origin and its limits. Where we do not have a source, we say so.

Where does the 23-minute refocus claim actually come from?#

The 23-minute figure — sometimes quoted with theatrical precision as 23 minutes and 15 seconds — traces to observational research by Gloria Mark and colleagues at UC Irvine, who followed information workers in their natural office environment and tracked how they moved between tasks throughout the day. When they measured the average time from an interruption back to the original task, accounting for all the intermediate activities in between, they arrived at a figure in that range.

That study is real and peer-reviewed. But the way it circulates strips out three things that change what it means. First, the sample: this was a small group of workers observed over a limited period. Small observational studies establish patterns; they do not establish universal constants. Second, what counts as recovery: the time measured includes everything that happened between the interruption and the eventual return to the original task — reading other emails, checking notifications, working on unrelated items. It is the average time before someone got back to what they were doing, not a measure of how long the brain takes to regain cognitive depth once they returned. Third, subsequent research by Mark found faster recovery times in different conditions, particularly for shorter tasks and simpler interruptions.

The honest read: the 23-minute figure is not invented, but it is not a universal biological constant either. It describes average observed behavior in specific workplaces under a specific methodology. Both the sample size and the definition of recovery should temper how confidently you apply it to your own inbox.

What does the broader task-switching research actually show?#

Separate from the observational study above, cognitive psychologists have run controlled laboratory experiments measuring what happens when people switch between tasks in real time. The American Psychological Association has summarized this line of research: when people switch back and forth between tasks, they take measurably longer to complete them than when they work on one task continuously. This time cost — the switching cost — comes from the cognitive overhead of loading a new task into working memory while the previous task is still partially active.

Research by Joshua Rubinstein, David Meyer, and Jeffrey Evans documented that even brief mental blocks created by task-switching add up cumulatively during complex work. The key finding is not a single number but a principle: the productivity cost of switching tasks is real, scales with task complexity, and is not simply proportional to the length of the interruption. You pay the switch cost on the way in and again when you return.

What this literature does not establish is a single universal interruption tax applicable to every worker in every context. Controlled experiments use specific tasks under laboratory conditions, which limits how directly the findings transfer to real offices where tasks vary enormously in complexity and familiarity. The principle is robust; specific numbers are not portable.

Common claims versus what the research actually supports#

What is attention residue, and why does it explain more than the 23-minute figure?#

Attention residue is a concept introduced by Sophie Leroy at the University of Minnesota to explain why task-switching carries costs that extend beyond the duration of the switch itself. The observation: when you stop working on task A to switch to task B, part of your cognitive attention remains allocated to task A. You are physically present with task B but mentally still partly engaged with what you left. Task B gets a compromised, partially occupied version of your attention — which slows it down and reduces quality.

This mechanism matters more than any headline number because it explains the shape of the problem without requiring a specific figure. A 30-second glance at a notification during deep work does not cost 30 seconds — it costs the attention residue it generates for the next several minutes of that focused work. The total cost is paid by the work you return to, not by the interruption itself. This also explains why the length of an interruption is a poor proxy for its cost: a brief interruption that hits when you are deep in a complex task carries high residue; a longer one that arrives between tasks carries far less.

The practical implication of the attention residue framework is that the research-supported intervention is not faster interruption handling but fewer interruptions during periods of high-complexity work. Checking email in two or three concentrated windows instead of continuously is not about the minutes saved — it is about keeping attention residue low during the blocks in between.

Does email specifically carry the same switching costs as other interruptions?#

Email amplifies the switching cost problem with a structural feature specific to its design. It is asynchronous communication presented synchronously: a notification arrives in real time for something that almost never required an immediate reply. That alert creates an interruption at whatever moment in your focus session it hits, regardless of the urgency of the message. The cost of the interruption is entirely unrelated to the importance of the email.

Email also typically demands context reconstruction in a way that many interruptions do not. A colleague interrupting you with a question about the thing you are actively working on carries relatively low switching cost — you are already in context. An incoming email usually requires you to rebuild an entire prior thread before you can evaluate or respond to it, which layers cognitive load on top of the basic switching cost.

The Microsoft Work Trend Index has tracked email and digital message volume over time, finding that knowledge workers are sending and receiving substantially more messages than they were five years ago. More messages means more potential interruption events and a higher baseline rate of context-switching throughout the working day. The research does not produce a clean per-email cost figure, but the mechanism it establishes — repeated switching imposes cumulative time costs — applies directly to a high-volume inbox monitored continuously.

How to reduce interruption costs based on what the research actually supports#

The practical takeaways from this literature are narrower than productivity advice usually makes them. The research supports a small number of concrete interventions with genuine mechanism backing, and many popular recommendations have weaker support than they are given credit for. These are the ones with the clearest grounding.

1. 1

   Close the inbox during focus blocks, not just silence it

   The notification is not the only source of interruption cost. Having an inbox tab open during focused work creates self-initiated interruptions as you check it out of habit. Closing the application entirely — not just muting notifications — is what the attention residue research supports. Out of sight does not eliminate the temptation, but it raises the friction enough to reduce self-interruption substantially.

2. 2

   Batch email into two or three defined windows per day

   Processing email in concentrated sessions rather than continuously is the most directly research-supported behavioral intervention. It converts a continuous source of random interruptions into a defined task with clear entry and exit, which means the attention residue from email processing is confined to the periods around those windows rather than distributed throughout the day. The number of windows matters less than the discipline of exiting completely between them.

3. 3

   Sort before you reply within each session

   During each inbox session, scan and categorize before composing any reply. Deciding what each message is — whether it needs your reply, someone else's, can wait, or can be ignored — is lower-complexity work than composing. Completing the sorting pass before the reply pass means you handle all the low-complexity switching in one block, rather than alternating between categorization and deep writing throughout the session.

4. 4

   Leave a re-entry note before switching away from a task

   Before switching away from a focus task — even for a planned email session — write one sentence describing exactly where you are and what the next action is. Switching costs are significantly lower when you can pick up from a clear written note rather than reconstructing your state from scratch. This is particularly valuable for complex writing or analysis that takes multiple sessions.

5. 5

   Remove peripheral signals, not just notification sounds

   The attention residue effect is triggered by any attention switch, including ones you initiate yourself. Disabling notification badges, email preview text, and unread counts removes the peripheral signals that prompt self-interruption during focus work. Badge counts on an inbox app are designed to make you look; the research-consistent intervention is removing the cue, not just the sound.

A faster way: handling email interruptions systematically#

The steps above work, and none of them require software. But maintaining those boundaries consistently — closing the inbox before each focus block, remembering to batch rather than monitor, rebuilding context on return — is itself a recurring tax on attention throughout the day. The discipline competes with the focus it is trying to protect.

We build AI Emaily, an AI-native email client, and the research above is the problem it was built around. When AI handles the low-urgency flow continuously — sorting noise from what matters as mail arrives, drafting routine replies for your approval, tracking threads that need follow-up — the inbox stops requiring active monitoring. You can leave it closed for a two-hour focus block knowing that anything genuinely urgent has been surfaced and the routine is in progress. The mechanism the research supports — fewer interruptions during high-complexity work — happens as a byproduct of the AI's work rather than a discipline you have to sustain manually. You can see how it works at aiemaily.com or review plan details at aiemaily.com/pricing.