It comes up constantly: a Slack message lands mid-task, someone mentions in a meeting that it now takes 23 minutes to get back on track, everyone nods. The number is almost always credited to "a UC Irvine study" or "research by Gloria Mark," rarely to a specific paper. We went and read the papers.
Where the Number Is Supposed to Come From
The figure traces to a 2006 interview Gloria Mark, a real UC Irvine researcher who has spent two decades studying interruptions, gave to Gallup Business Journal. In it, she says interrupted work "was resumed, on average, in 23 minutes and 15 seconds." That's the origin most people are actually quoting, whether they know it or not: an interview statement, not a citation.
What the Studies Actually Found
Mark's two most-cited papers on this subject say something different. A 2005 study observed 24 real information workers, in detail, over roughly 700 hours across thirteen months at a US financial-services company. It found people spend about 11 minutes in a given task before switching to another, that 57% of these stretches get interrupted, and that when interrupted work does get picked back up, more than two other activities usually happen first. There's no resumption-time-in-minutes figure anywhere in it.
A 2008 study, a separate lab experiment with 48 German university students doing a simulated email task, found something closer to the opposite of the popular story: people who weren't interrupted at all took the longest to finish. The interrupted groups worked faster, with no drop in the quality of their work, but reported significantly more stress, frustration, time pressure, and effort. Again, no 23-minute figure appears anywhere in it.
Not Fake, Just Not There
None of this means interruptions are free, or that Mark's research is wrong. The 2008 study is fairly clear that interruptions cost something, just not the thing everyone assumes. What we can say, having actually read the two papers this number is usually pinned to, is that it isn't in either one. Whether it appears somewhere else in her broader body of work, we can't rule out entirely. What we can say is that the version repeated everywhere, credited to "a study," traces to an interview quote instead.
Both studies also come with real limits worth naming rather than glossing over: the 2005 study is one company, not a cross-section of office work; the 2008 study is German university students in a lab, not a UK desk job. Neither is proof of a universal number, which is exactly why treating either as one would repeat the same mistake this piece is pointing at.
Why 0800Labs Is Interested in This Moment
We didn't go looking for a number to debunk for its own sake. What interests us is smaller: how often a workday fact gets repeated so widely that nobody thinks to check it, the same pattern that shows up in the "87% of workers hit a 3pm slump" claim we traced and couldn't verify. Both are stand-ins for something real (people do get tired at 3pm; interruptions do cost something) wearing a precision the evidence doesn't actually support.
If that kind of honesty is useful, the waitlist is below.