What Does Gloria Mark's Research Say About Interruptions at Work?
Written By Aftertone Team
Thursday, May 14, 2026
15 min read
What Does Gloria Mark's Research Say About Interruptions at Work?
Gloria Mark's research on workplace interruptions is among the most-cited findings in productivity science. Her field studies at UC Irvine, tracking knowledge workers at second-by-second resolution, produced the finding that it takes an average of 23 minutes and 15 seconds to return to the original task after an interruption. But the research is richer and more specific than the headline figure suggests, and understanding what it actually measured changes how the finding should be applied.
The core studies
Mark's most influential work came from a series of naturalistic field studies tracking knowledge workers in their actual work environments rather than in laboratory settings. The 2008 study with Daniela Gudith and Ulrich Klocke tracked 36 workers at second-by-second resolution across their working days, recording every task transition, every interruption, and every return to prior tasks. The study produced several findings that are individually important.
The 23-minute-15-second figure describes the average time before workers resumed their original task after an externally imposed interruption. But the figure conceals a crucial detail: workers did not return directly. They worked through an average of two other tasks before returning to the interrupted task. The 23 minutes is not 23 minutes of idle recovery. It is 23 minutes of apparent productivity during which the interrupted task's context is not yet restored.
A second finding from the same study: interrupted work was resumed on the same day 81.9% of the time. The interruption did not kill the task in most cases. It delayed it, but the delay was costly in the ways described above rather than in the blunter way of complete task abandonment.
Mark's research also found that the average duration of uninterrupted work on a single task was approximately 3 minutes 5 seconds. This figure, from her 2004 study with Victor Gonzalez, was even more striking than the recovery time: workers were switching contexts every three minutes on average, making sustained deep work essentially structurally impossible in the environments studied.
Self-interruptions versus external interruptions
Mark and colleagues' research distinguished between externally imposed interruptions (a colleague approaching, a notification arriving, an urgent request appearing) and self-initiated interruptions (the worker choosing to switch tasks, check email, or attend to something else). Both types were prevalent; self-interruptions were not rare. Workers interrupted themselves nearly as often as they were interrupted by external sources.
This finding is practically significant. It means that eliminating all external interruptions would not eliminate the problem. A significant portion of the context switching that degrades knowledge worker performance is self-generated, reflecting the cognitive response to the low-engagement periods in any task rather than purely environmental noise. The intervention required is both environmental (reducing external interruption sources) and behavioural (reducing self-interruption tendencies, which is a function of attentional training and environmental design that removes accessible distraction alternatives).
Later research: notifications and digital interruption
Mark's subsequent research extended to digital environments. A 2016 study found that the average person checks their email 77 times per day and their phone 85 times per day, with each check representing a context switch that generates the switching and residue costs her earlier research identified. The cumulative switching cost across a workday of this density is substantial.
In her book Attention Span (2023), Mark synthesised her decades of research with a finding that has attracted particular attention: average attention span on screens has dropped significantly over the 15 years she has been measuring it, from 2.5 minutes in 2004 to around 47 seconds in more recent measurements. This is a measured average from her specific research cohort and methodology, not a claim about all attention in all contexts, but it reflects the increasingly fractured attentional environment that her earlier work predicted would develop.
What the research does and does not show
Mark's research shows that externally imposed interruptions to knowledge work generate a recovery cost that extends well beyond the duration of the interruption, typically 20 to 25 minutes for full return to depth. It shows that self-interruptions are common and contribute significantly to context switching costs. It shows that the average frequency of task switching in observed knowledge work environments is far higher than most workers realise or would choose if they were tracking it.
The research does not show that 23 minutes is a fixed biological constant. The figure is a population average from specific study populations in specific environments. Recovery time varies with interruption type, task complexity, completeness of the interrupted task, and the quality of the environment during recovery. The direction (recovery takes significantly longer than the interruption) is well-established; the exact duration is context-dependent.
The research also does not show that all context switching is equally costly. Switching between very similar tasks, switching at natural stopping points, and switching between tasks with low working memory demands generates less residue and shorter recovery times than switching from demanding tasks in mid-flow. The 23-minute figure represents the worst-case type of interruption: an external interruption to complex work during active engagement.
Aftertone's planned versus actual report surfaces the interruption cost that Mark's research quantifies: the gap between what was planned for a day and what was completed reflects the aggregate switching cost and residue from actual interruption density. Two to three weeks of tracking makes the relationship between meeting load and deep work output concrete rather than intuitive.
Frequently asked questions
What is Gloria Mark's research on interruptions?
A series of field studies tracking knowledge workers at second-by-second resolution in their actual work environments. The most cited finding is from a 2008 study with Gudith and Klocke: after an external interruption, it takes an average of 23 minutes and 15 seconds before workers return to the original task, typically via two intervening tasks. A 2004 study with Gonzalez found average uninterrupted work time on a single task was approximately 3 minutes 5 seconds.
Is the 23-minute recovery time a real finding?
The 23-minute recovery time is a real finding from a specific field study, but it is a population average rather than a fixed biological constant. The figure comes from Gloria Mark, Gudith, and Klocke's 2008 UC Irvine study tracking 36 workers at second-by-second resolution. Recovery time varies based on interruption type, task complexity, and environmental conditions during recovery. The direction is well-established: recovery from external interruptions to complex knowledge work takes significantly longer than the interruption itself.
How many interruptions does the average knowledge worker experience per day?
Mark's research found approximately 275 interruptions per day on average in her study populations, including both external interruptions and self-initiated task switches. At this frequency, full 23-minute recovery between interruptions is mathematically impossible, which explains why most knowledge workers operate in a state of perpetually incomplete recovery throughout the workday.
Did Gloria Mark find that attention spans are shrinking?
Her research found that average screen attention span in her study cohort decreased from approximately 2.5 minutes in 2004 to approximately 47 seconds in more recent measurements. This is from her specific research methodology and cohort; it should not be generalised to all attention in all contexts. It reflects the increasingly fragmented attentional environment of digital knowledge work rather than a universal biological change in human attention capacity.
What does Gloria Mark's research mean practically for productivity?
That the cost of interruptions to knowledge work is substantially higher than their duration suggests, that self-interruptions are nearly as costly as external ones, that context switching is far more frequent in typical work environments than most workers realise, and that creating environments with fewer interruption sources and fewer accessible distraction alternatives is more effective than attempting to maintain focus through willpower in an interrupt-rich environment.
