Forty-seven seconds: how short your attention span has become

The headline number in the modern science of attention is small enough to fit on a wristwatch.

Forty-seven seconds.

That is the average uninterrupted stretch of focused attention an office worker spends on a single task at a computer before switching to something else, according to the longest-running study of the question we have. It used to be 150 seconds. It is not a meme, it is not a tabloid stat, and it is not the goldfish-comparison the internet likes to circulate. It is real, measured data, gathered over twenty years by one researcher.

This piece is about who she is, how she measured it, what the number actually means, and what to do about it.

What is the attention span of humans?

Forty-seven seconds, on average, is how long people focused on a single task at a computer screen before switching, in measurements taken in 2021.^[2] That is an average across many stretches of screen-based work, not a hard limit. Many stretches are longer and some much shorter, and it measures the gap between switches rather than the longest anyone can concentrate.

How short is the average attention span?

About 47 seconds on a screen, and the typical stretch is shorter than that. Gloria Mark reports that the median is 40 seconds: half of all the focus stretches her team logged were under 40 seconds before the person switched to something else.^[8] When her measurements began in 2004 the average was about two and a half minutes.^[1][8] The time was logged automatically in the background while people did their normal work, so it is a measure of how often we switch on screens, not a limit on how long anyone can concentrate on a film, a book or a conversation.

Is attention span decreasing?

Yes, on this measure. The average stretch was about 150 seconds in 2004, 75 seconds by 2012 and 47 seconds by 2021.^[1][2] The figures come from office workers at computers, so they describe attention during screen-based work rather than every kind of attention.

Why is attention span decreasing?

Two things sit behind the fall: an environment built to interrupt us, and a trained habit of interrupting ourselves. In 2004 almost all interruptions in the studies were external, such as a notification, a colleague or a phone ringing. By 2021 a substantial share were self-initiated: switching tabs, refreshing a feed, checking the phone by the keyboard.

Can attention span be improved?

Yes, with consistent practice rather than one big fix. The best-supported moves are environmental: move the phone to another room, batch notifications, and practise sustained-attention activities such as reading on paper or walking without audio. The deep-attention circuit responds to use within weeks, though twenty years of trained habit will not reverse in a single session.

What does a short attention span mean?

It means finding it hard to stay with one thing before your attention jumps to something else. In the research on this page, that is measured very concretely: the number of seconds someone stays on one screen task before switching, which averaged 47 seconds in 2021.^[2] A short attention span in that sense describes a habit of switching, shaped by your surroundings, rather than a fixed limit on how long you are able to concentrate.

How does a short attention span affect learning?

10.6% lower test scores was the cost in one classroom experiment. Students watched a 30-minute recorded lecture while receiving text messages that asked for replies, and those who sent and received 16 or more texts scored 10.6% lower on a recall test afterwards than those who texted little or not at all.^[4] There was no significant difference between the low and moderate texting groups, so the damage came from heavy switching rather than the odd interruption. Students who left a longer gap before replying also did better, which is the practical lesson: the fewer times attention is pulled away, the more of the lesson stays.

Can you fix a short attention span from TikTok?

Yes, and faster than most people expect. In a 2025 randomised trial, people who had mobile internet blocked on their phones for two weeks improved on an objective test of sustained attention, by an amount the researchers compared to about ten years of age-related decline.^[6] The study blocked mobile internet in general rather than TikTok specifically, and participants could still call, text and use a computer. Nothing here is medical advice, but the direction is clear: less short-form input on the phone, and attention starts to come back within weeks.

Do goldfish really have a longer attention span than us?

No. The claim that people now have an 8-second attention span, shorter than a goldfish’s 9 seconds, was traced by the BBC’s More or Less programme in 2017. It led back to a 2015 report by Microsoft Canada’s consumer insights team, but the figure did not come from Microsoft’s own research: the report cited a website called Statistic Brain, and the sources that site listed could find no research behind the numbers.^[7]

Attention researchers interviewed for the piece said an “average attention span” is not really something psychologists measure, because attention depends on the task. And goldfish are not the forgetful creatures of the joke: they have long been used as a model species for studying learning and memory.^[7] The 47-second figure on this page is a different thing, a real measurement of how often people switch between tasks on a screen.

Who Gloria Mark is

Dr Gloria Mark is a professor of informatics at the University of California, Irvine, and one of the most-cited researchers in the field of human-computer interaction. Since the early 2000s she has been measuring something almost everyone else in technology research has ignored: how long the average person actually focuses on a single thing at a screen, in their real working life, before switching.

The methodology, simplified: a study participant goes about their normal work at a normal screen. Their activity is logged. Every time they switch from one application or document to another, a switch event is recorded. The duration between switches is the “attention stretch.” Average it across thousands of stretches and you get a figure for that population in that period.

In her 2004 work she reported an average of approximately 150 seconds.^[1]

In her 2008 follow-up the figure had fallen.^[3]

In her 2023 book, Attention Span, she presented the longitudinal trajectory in full. By 2012 the average had reached 75 seconds. By 2021 it was 47 seconds.^[2]

What the number does and does not say

The 47-second figure does not say that no human in 2021 could focus for longer than 47 seconds. It is an average across stretches. Many stretches are longer. Some are much shorter. What the number captures is the typical gap between context switches in modern computer-based work.

The decline is striking on its own. What is more striking is what changed alongside it. In 2004, almost all interruptions in the studies were external: a notification, a colleague, a phone ringing. By 2021, a substantial proportion of interruptions were self-initiated: the worker switching tabs without external prompting, checking something, refreshing a feed, checking the phone next to the keyboard.

This is the key point about modern attention erosion. The trained behaviour is now internal. We have become our own interrupters. The smartphone is not in the room any more; the smartphone has been internalised.

Why it matters

A 47-second attention stretch is not enough time to think a complete long thought. It is not enough time to settle into a difficult passage of text, to construct a multi-step argument, to plan a complex piece of work, or to hold a long question in mind while looking for the answer.

The research on the cost of constant context-switching is consistent. Larry Rosen and colleagues at California State University have shown that switching attention to a notification or a brief text message imposes a measurable cost on the task one returns to: comprehension drops, error rates rise, completion times extend.^[4] The effect compounds with frequency.

There is also a wellbeing dimension. Matt Killingsworth and Daniel Gilbert ran a large-scale experience-sampling study in 2010 using an app that pinged participants at random moments asking what they were doing and how they felt.^[5] They collected over 250,000 data points across thousands of participants. The headline finding was simple: a wandering mind is an unhappy mind. People reported lower wellbeing when their attention was scattered than when it was settled, regardless of what they were actually doing.

(There is a subtle but important caveat to that study: not all mind-wandering is the same. Productive mind-wandering, of the default-network kind that generates ideas, is different from the anxious, distracted, jumping-between-things wandering that the phone trains. The Killingsworth result is mostly about the second kind.)

What to do

There is no silver bullet for reversing twenty years of trained attentional erosion in a single intervention. There is, however, a small number of consistent, evidence-supported moves:

• Remove the most-frequent self-interrupter from the environment. For most people, this is the phone in the same room as the work. Move it. Do not “put it on silent.” Move it to another room. The internal pull weakens within days.
• Batch notifications. Almost no notification, including work ones, actually needs to be delivered the second it arrives. Setting “do not disturb” windows of even one hour at a time triples the average attention stretch in studies on knowledge work.
• Practise sustained-attention activities deliberately. Reading on paper. Long walks with no audio input. Writing by hand. The deep-attention circuit responds to use within weeks.
• Stop using “five-minute breaks” to scroll. The instinct is to think the break is recovery. It is not. The scroll keeps the attentional system in interruption mode. Genuine recovery is doing nothing for the five minutes.

The 47-second figure is not destiny. It is an aggregate of an environment we have built that interrupts us, plus a trained behaviour where we have learned to interrupt ourselves. Both of those can be changed, and the data on rehabilitation is encouraging.

For the longer argument, the small book this studio publishes, How to Be Bored Again, sits at the centre of this question. The first half of the book is the diagnosis, the second half is the practice. Twelve short chapters. About a long evening's read.

There is a small, almost humorous test of where your own attention is right now: try reading the previous paragraph of this article again, without checking anything else, and see whether you make it. If you did, you have already begun.