Your Attention Span Is Now 47 Seconds. Here's What That Means for Your Workday
Gloria Mark's decades of research put the average screen attention span at 47 seconds. Here's what that number actually means, and how it should change the way you structure a workday.

Your Attention Span Is Now 47 Seconds. Here's What That Means for Your Workday
Gloria Mark has spent nearly two decades studying how people actually use technology, not in a lab with a fixed task, but in what she calls "living laboratories": real offices, real inboxes, real days. Her research, summarized on her own site ahead of her book Attention Span, produced a number that's become one of the most cited (and most misunderstood) statistics in workplace productivity: the average attention span on any given screen is now just 47 seconds.
That number alone tends to get used as a scare stat. Mark's actual point is more useful than that, and more specific to how a workday should be structured.
The Number Isn't the Whole Story
Forty-seven seconds describes how long, on average, someone stays on a single screen before switching, not how long someone can focus if they try. Mark's research draws a clear line between attention spans shrinking and people being "doomed to live unfocused lives." Her framing is that the way we focus is changing, not disappearing, and that trying to force old-model, unbroken concentration onto a workday built around constant digital interruption is a losing strategy regardless of willpower.
Myth One: You Should Always Be Trying to Focus
The first myth Mark's research directly challenges is the idea that sustained, uninterrupted focus is both possible and something to feel guilty about failing at. Her comparison is physical: you can't lift weights nonstop for hours, and you can't hold sustained cognitive focus for long stretches either without replenishing the resource. Her research identifies different types of attention, some that tax you the way a hard set does, some that genuinely restore you the way rest between sets does, and argues those types follow a rhythm. The practical implication is that scheduling a workday around rigid, back-to-back deep-focus blocks with no recovery built in works against how attention actually functions, not with it.
Myth Two: "Wasted" Downtime Is Actually Wasted
The second myth is that low-effort digital activity, the mindless scrolling or a quick simple game between tasks, is pure waste. Mark's research found that used strategically, these activities can replenish overspent mental resources and even help fresh ideas surface. That doesn't mean unlimited scrolling is good for output, but it reframes the five-minute break between meetings as a potential recovery mechanism rather than automatically a discipline failure.
Myth Three: It's Just a Notifications and Willpower Problem
The third myth is that distraction is mostly self-inflicted, too many notifications, not enough discipline. Mark's work points to broader sociotechnical forces: algorithmic design, the ambient exposure to fast-cut media across the day, and even individual personality differences in susceptibility to distraction. That matters for how organizations think about the problem. A notification-silencing policy addresses one input among several, not the underlying design of how digital work is structured.
Myth Four: Flow Is the Goal
The last myth Mark takes on is the popularized idea that "flow," a state of total immersive absorption, is the ideal state every knowledge worker should be chasing. Her argument is that flow isn't realistic for most jobs, which require an analytical, evaluative mindset rather than pure immersive execution. The healthier goal, in her framing, is a balance across different attentional states rather than a permanent hunt for one specific, hard-to-sustain one.
Where AI Agents Fit Into a 47-Second Attention Span
If the workday runs on short, constantly interrupted bursts of attention rather than long unbroken stretches, that has a direct implication for how AI agents get used. An agent doesn't need one person's continuous focus to be useful; it can keep working in the background across the same gaps and context-switches that already fragment a human day. That's a different fit than a tool that demands sustained attention to operate.
In practice, this plays out as running several agents in parallel: one drafting a follow-up email while another is pulling numbers into a report, a third monitoring a shared inbox for anything urgent, all progressing during the same 47-second windows a person spends glancing at a phone or switching tabs. The person checks back in, reviews what moved forward, and redirects the next piece, rather than personally holding every task open in their head between checks. That maps onto Mark's point about attention working in rhythms, not a single depletable tank: agents can occupy the "in-between" moments a fragmented day already produces, instead of competing with a person for the same scarce, sustained-focus time.
The efficiency gain isn't from getting a human to focus harder or longer. It's from letting the software absorb the parts of the work that would otherwise require exactly the kind of continuous attention Mark's research says is increasingly unrealistic, so the person's actual attention gets spent on the review-and-decide moments instead of the execution moments.
What This Means for How You Structure a Workday
None of this is an argument against focus, it's an argument against structuring a workday as if attention were a single, depletable tank you either protect perfectly or waste. Mark's research suggests a more realistic model: workdays that build in actual recovery between demanding tasks, that don't treat every short mental break as evidence of poor discipline, and that account for the fact that the modern information environment, not just individual willpower, shapes how attention holds up over a day. Teams and individuals who plan around that rhythm, rather than against it, are working with the grain of how attention actually behaves instead of against it.
Frequently Asked Questions
Is the 47-second attention span statistic about how long someone can focus if they try hard? No. It measures average time spent on a single screen before switching, based on Gloria Mark's long-running observational research, not the ceiling on sustained concentration under effort.
Does this mean people's ability to focus is permanently declining? Mark's research frames it as a change in how attention operates in a technology-saturated environment, not an irreversible decline. She argues the goal should shift from chasing constant focus to working with natural attentional rhythms.
Is taking a break during a demanding task actually bad for productivity? Not according to this research. Mark's work found that even low-effort activity, used strategically, can replenish mental resources depleted by focused work rather than simply wasting time.
Should teams still aim for "flow" states during deep work? Mark's research questions this as a universal goal, arguing that most jobs require an analytical mindset that doesn't fit the immersive-flow model, and that a balance of attentional states is a more realistic target than flow itself.
Workmate
See what an agent team would do for your business.
Keep reading

Why Most Private Equity Firms Aren't Seeing Returns From Their AI Yet
88% of PE firms have put $1M+ into GenAI for dealmaking, and most portfolio companies have deployed AI tools. BCG's research says almost none of that has translated into real returns yet, and explains exactly why.

AI Agents for Insurance Teams: Underwriting, Claims, and the Coverage Gap
82% of carriers are planning agentic AI adoption within three years, per Deloitte. Here's where AI agents are landing in underwriting, distribution, and the industry's persistent coverage gap.

AI Agents for Healthcare Teams: Scheduling, Billing, and Follow-Up Care
Half of US healthcare orgs have now implemented gen AI, and agentic AI is next in line for the scheduling, billing, and follow-up work that's been stuck in staffing shortages for years. Here's what McKinsey's own late-2025 survey data shows is actually working.