Attention Span Statistics 2026: 90+ Facts and 12 Myths

Attention span statistics are some of the most quoted numbers in marketing, and some of the least checked. The most famous of them, that humans now have an 8-second attention span, shorter than a goldfish's, has no study behind it at all. We traced it back to its source, and the trail ends in a 2015 slide that cites a website that cited nobody.

The real research is more useful than the myth. People switch screens every 47 seconds at work, social feed ads get around 2 seconds of real attention, and short videos are watched to the end far more often than long ones. This page collects more than 90 of those findings from peer-reviewed studies, regulators and platforms, checked against each publisher in October 2026, plus 12 popular claims that do not hold up. The period and source are named on every line, and vendor data is labelled as such.

Attention span statistics at a glance

• 47 seconds: average time knowledge workers stay on one screen before switching, down from 2.5 minutes in 2004 (Gloria Mark, UC Irvine)
• 8 seconds: the famous "human attention span" figure, which has no primary source (BBC More or Less, 2017)
• 1.7 seconds: average time people spend with a piece of content in the Facebook mobile feed (Meta, 2016)
• 2.4 to 2.8 seconds: active attention per social feed ad (Amplified, 2026, modelled)
• 65% of viewers finish business videos under one minute, against 20% for videos over 20 minutes (Vidyard)
• 186 phone checks a day for the average American (Reviews.org, 2026, self-reported)
• r = -0.38: correlation between heavier short-form video use and poorer attention across 71 studies (Psychological Bulletin, 2025)
• 20% to 28%: share of words on a web page users have time to read (Nielsen Norman Group, 2008)

The 8-second goldfish attention span myth

The claim usually reads: the average human attention span fell from 12 seconds in 2000 to 8 seconds in 2013, one second shorter than a goldfish.

It spread through a 2015 Microsoft Canada consumer insights report on attention spans. The relevant slide credits its numbers to "Statistic Brain", a statistics website. Statistic Brain's page, in turn, listed the US National Library of Medicine and the Associated Press as sources. When BBC Radio 4's More or Less investigated in 2017, neither organisation could find any research behind the figures.

Three more problems with it:

1. Microsoft's own research, a survey of 2,000 Canadians plus EEG tests, never measured an attention span in seconds. The 8-second figure sits on a slide sourced elsewhere.
2. "Average attention span" is not a standard measure in psychology. Attention varies by task, motivation and context, so a single number for all humans means little.
3. Goldfish do not have short memories. Fish behaviour researchers use them as a model for studying learning precisely because they remember things for days and weeks; a 1994 study trained goldfish to press a lever for food at a set time of day.

The original Microsoft report is now only available through the Internet Archive. Its real findings are more modest:

4. 44% of Canadians said they had to concentrate hard to stay focused on tasks, rising to 67% of 18 to 24 year olds (2014 survey of 2,000 people).
5. The report itself states that age is not significantly tied to sustained attention.

Attention span statistics at work: screens and interruptions

The most cited real research comes from Gloria Mark at UC Irvine, who has measured how long people stay on one screen or task before switching since 2004. Her figures measure switching behaviour among knowledge workers, not attention capacity, and her methods moved from stopwatch observation to software logging around 2011.

From Mark's interviews, including a 2024 Hidden Brain episode:

6. In 2004, average attention on one screen was about 2.5 minutes before switching.
7. By 2012 it had fallen to about 75 seconds.
8. From about 2016 to 2020 it averaged 47 seconds, with other labs finding 44 to 50 seconds.

From Mark's earlier field studies, including her 2005 CHI paper and a 2006 Gallup interview:

9. Information workers spent an average of 11 minutes 4 seconds on a project before switching or being interrupted (2005, 24 workers, 700+ hours observed).
10. 57% of these working spheres were interrupted.
11. Interrupted work resumed the same day took an average of 25 minutes 26 seconds to get back to, with about 2.26 other tasks in between.
12. Workers juggled 12.2 different projects a day and switched between them about every 10.5 minutes (2006 interview, 36 workers).
13. About 44% of interruptions were self-inflicted rather than external.
14. Workers spent about 3 minutes 5 seconds on any single event before switching, and 2 minutes 11 seconds on any one device (2004 observation).

On email and notifications:

15. Information workers checked email 77 times a day and spent almost 1.5 hours a day on it (Microsoft Research and UC Irvine, 2016, 40 workers).
16. The longer people spent on email in a day, the lower their perceived productivity and the higher their measured stress (same study).
17. After switching to an email alert, people took an average of 9 minutes 33 seconds to return to the application they had been using (Microsoft Research, 2007, 27 employees).
18. 27% of task suspensions caused by alerts lasted more than two hours (same study).
19. Phone notifications alone, without touching the phone, significantly disrupted performance on an attention task, comparable to using the phone (Florida State University, 2015).
20. Heavily messaged Microsoft 365 users are interrupted every 2 minutes during core hours, or 275 times a day, according to Microsoft's 2025 Work Trend Index. This covers the top 20% most-pinged users, not the average employee.
21. 48% of employees and 52% of leaders say their work feels chaotic and fragmented (Microsoft, 2025 survey of 31,000 people).

Multitasking statistics

22. Heavy media multitaskers performed worse at task switching and were more distracted by irrelevant information than light multitaskers (Stanford, PNAS, 2009).
23. A 2018 PNAS review found heavier media multitaskers tend to do worse cognitively, though many studies find no difference and causality is unknown.
24. Heavier media multitasking was linked to more attention lapses and more forgetting in an EEG study of 80 young adults (Stanford, Nature, 2020).
25. Switching between tasks carries measurable time costs that grow with task complexity (Journal of Experimental Psychology, 2001).
26. Having a smartphone on the desk, even switched off and face down, reduced available cognitive capacity compared with leaving it in another room, across nearly 800 people (University of Texas at Austin, 2017).
27. A 2023 meta-analysis of 22 studies found an overall negative effect of smartphone use and presence on cognitive performance, though not every cognitive area was affected equally.

Phone and screen time statistics

Phone check surveys are self-reported and run by companies that sell phone products, so treat the exact numbers loosely:

28. Americans say they check their phones 186 times a day, or 11.6 times an hour, according to Reviews.org's 2026 survey of about 1,000 adults.
29. 84.6% check their phone within 10 minutes of waking (Reviews.org).
30. 45.8% consider themselves addicted to their phone, a self-applied label rather than a clinical one (Reviews.org).
31. A 2019 Asurion survey of 1,998 people found 96 checks a day, once every 10 minutes. The two surveys used different methods, so the rise is not a like-for-like comparison.
32. 18 to 24 year olds checked their phones about twice as often as the US average (Asurion, 2019).

From government and research bodies:

33. 50.4% of US teens aged 12 to 17 had 4 or more hours of daily screen time outside schoolwork (CDC National Center for Health Statistics, 2021 to 2023).
34. Among those teens, 27.1% reported anxiety symptoms and 25.9% depression symptoms. This is an association, not proof of cause.
35. US tweens aged 8 to 12 used 5 hours 33 minutes of entertainment screen media a day, and teens 8 hours 39 minutes (Common Sense Media, 2021).
36. Children aged 0 to 8 average about 2.5 hours of screen time a day, with short-form video rising (Common Sense Media, 2025).
37. Nearly half of US teens say they are online almost constantly, up from 24% a decade earlier (Pew Research Center, 2024).
38. 16% of US teens say they are on TikTok almost constantly (Pew, 2024).
39. UK adults spent 4 hours 30 minutes online a day in May 2025, and 18 to 24 year olds 6 hours 20 minutes (Ofcom Online Nation 2025).
40. Smartphones account for 77% of UK adults' online time (Ofcom).
41. Watching TV remained Americans' top leisure activity at 2.6 hours a day in 2025 (US Bureau of Labor Statistics).

Reading and web attention statistics

From Nielsen Norman Group, whose figures are old but still the most cited on web reading:

42. On an average web page, users have time to read at most 28% of the words, and 20% is more likely (2008, based on 2005 browsing data from 25 users).
43. Users spend only 4.4 seconds more for each additional 100 words on a page.
44. 17% of page views lasted less than 4 seconds.
45. In NN/g's 1997 study, 79% of test users always scanned new pages and only 16% read word by word.
46. Writing that was concise, scannable and objective improved measured usability by 124% in the same 1997 test.
47. The F-shaped scanning pattern first identified in 2006 still appeared on desktop and mobile in 2017 eye tracking, though NN/g notes good design can prevent it.

On collective attention, from a 2019 Technical University of Denmark study in Nature Communications:

48. A top-50 Twitter hashtag stayed in the top 50 for 17.5 hours on average in 2013, falling to 11.9 hours in 2016, based on about 43 billion tweets.
49. The researchers stress this measures how quickly public topics turn over, and that the study does not address individual attention spans.

Short-form video and attention

50. A meta-analysis of 71 studies and 98,299 people found heavier short-form video use is associated with poorer attention (r = -0.38) and poorer impulse control (r = -0.41) (Psychological Bulletin, 2025). The studies are correlational, so they cannot show short videos cause the difference.
51. The same analysis linked short-form video use with poorer mental health, especially stress and anxiety, but found no link with body image or self-esteem.
52. In a lab experiment with 60 people, scrolling TikTok, but not Twitter or YouTube, significantly reduced people's ability to remember planned tasks (LMU Munich, CHI 2023). It was a short-term, single-study effect.
53. Sustained attention ability peaks in the early 40s and then gradually declines, in a study of 10,430 people (Psychological Science, 2015).

For the wider picture on the format, see our short-form video statistics.

Attention span statistics for video and social content

From Meta's research on Facebook and Instagram feeds:

54. People can recall mobile feed content after just 0.25 seconds of exposure (Meta, 2016). That is recall at a measurable rate, not strong memory.
55. People spend 1.7 seconds on average with a piece of content in the mobile feed, against 2.5 seconds on desktop (Meta, 2016).
56. Up to 47% of a video campaign's value is delivered in the first 3 seconds, and up to 74% in the first 10 seconds (Meta with Nielsen, mid 2010s).
57. 65% of people who watch the first 3 seconds of a Facebook video watch at least 10 seconds, and 45% watch 30 seconds (Meta, mid 2010s).

From TikTok's own creative research:

58. 90% of TikTok ad recall impact is captured within the first six seconds (TikTok for Business, 2020 analysis).
59. Over 63% of the highest click-through TikTok ads put their key message or product in the first 3 seconds (TikTok, 2020).
60. Vertical TikToks have an average 25% higher 6-second watch-through rate (TikTok, 2020).
61. TikTok's current guidance is to introduce the proposition in the first 3 seconds and prioritise the hook in the first 6 seconds (2025).

On video length and completion:

62. 65% of viewers watch business videos under one minute to the end, against 20% for videos over 20 minutes (Vidyard, 943,305 videos in 2024, mostly B2B sales videos).
63. Videos under a minute average a 52% engagement rate, meaning viewers watch about half on average (Wistia, 13 million+ videos).
64. Viewers typically watch over half of educational and tutorial videos between 1 and 5 minutes long (Wistia).
65. Engagement drops 11% once business videos pass 30 minutes (Wistia).
66. Shorter videos have higher engagement rates, but a 10-minute video still earns more total watch time than a 1-minute one (Wistia, 2026).
67. Webinar replays of 31 to 45 minutes get more than twice the engagement of those under 30 minutes (Wistia, 2026).
68. In the first 2% of a video, engagement drops an average of 4.9% for 1 to 2 minute videos but 17.3% for 5 to 10 minute videos (Wistia, older analysis).

On viewer and marketer preferences:

69. 71% of marketers believe videos between 30 seconds and 2 minutes are the most effective length (Wyzowl, 2026 survey of 266 respondents).
70. 63% of consumers would most like to learn about a product by watching a short video (Wyzowl).
71. 60% of TikTok users say short-form video under 60 seconds is the format they interact with most (Sprout Social, 2026).
72. 52% of social users gravitate to short-form video on Instagram, and only 19% favour long-form (Sprout Social).
73. On YouTube, 52% prefer short-form but 49% still engage heavily with long-form (Sprout Social).
74. Short-form video is marketers' top ROI format at 49%, ahead of long-form at 29% (HubSpot, 2026).

On sound and captions:

75. Over 75% of Instagram Reels views are sound on (Meta, 2024).
76. Captioned Facebook video ads increase view time by 12% on average (Meta, mid 2010s).
77. 41% of Facebook video ads were basically meaningless without sound (Meta, mid 2010s).
78. 88% of TikTok users say sound is vital to the TikTok experience (TikTok with Kantar, 2022).

Advertising attention statistics

Attention measurement companies track whether ads are actually looked at, not just shown. Their figures come from eye tracking panels or models, so they vary by method.

79. Social feed ads earn just 2.4 to 2.8 seconds of active attention per impression, about 21% to 24% of their length, according to Amplified's 2026 study. It used AI-simulated viewers, covered Australia only and was run with Snapchat, which scored highest at 7.8 seconds.
80. About two thirds of viewers disengage within the first three seconds of a social feed ad, and only 8% to 10% remain by the end (Amplified, 2026).
81. Time in view overstates active viewing in social feeds by 1.5 to 2.5 times (Amplified, 2026).
82. Ads need about 2.5 seconds of active attention before memory effects become lasting, and over 80% of ads fall short of that threshold (Amplified Intelligence, 2022).
83. Well-branded ads can drive outcomes with just 1.5 seconds of attention (Amplified with VCCP Media, 2025).
84. Facebook-type video ads get a mean of about 1.7 seconds of attention, and most are merely glanced at (Lumen Research, 2022).
85. A 30-second US TV ad is looked at for about 13.8 seconds, and only 9% of TV ad impressions are watched all the way through (Lumen citing TVision data).
86. A 30-second UK TV ad receives 11.8 seconds of attention on average (Lumen, 2022).
87. Desktop display ads get 1.3 seconds of viewing, mobile ads 1.5 seconds and desktop pre-roll 9 seconds (Lumen, 2019).
88. Brand recall rose from 21% for ads viewed under one second to 41% for ads viewed over 10 seconds, in a Japanese study of 8,000+ mobile users (Dentsu, 2024).
89. People who chose to keep watching an ad for 10+ seconds recalled it 79% more than people forced to watch (Dentsu Japan, 2024).
90. Returns from attention to video ads begin to diminish beyond about 20 seconds of active viewing (Dentsu with Kantar and Lumen, 2026).

The industry standards behind "viewable":

91. A display ad counts as viewable when 50% of its pixels are on screen for at least one second, and a video ad after two continuous seconds (Media Rating Council, 2015).
92. The IAB and MRC published their first Attention Measurement Guidelines in November 2025, separating whether an ad could be seen from whether it was.
93. North American open-web video ads were 88% viewable and display ads 75% in Q1 2026 (DoubleVerify).
94. Average email read time fell to 8.97 seconds in 2022, from 13.4 seconds in 2018 (Litmus). Apple's 2021 privacy changes affect this data, and 2022 is the latest year published.

Attention span myths we checked

1. "Humans have an 8-second attention span, shorter than a goldfish." No primary source. It traces to a 2015 Microsoft Canada slide citing Statistic Brain, which cited organisations that could not find the research. Debunked.

2. "Goldfish have a 3-second memory." False. Goldfish are used to study memory and learning because they remember for days to weeks.

3. "Attention spans are shrinking every year." Misleading. People switch screens more often and public topics turn over faster, but there is no robust evidence that underlying attention capacity is falling year on year.

4. "It takes 23 minutes to refocus after an interruption." Real number, wrong meaning. Gloria Mark described an average of 23 minutes 15 seconds in a 2006 Gallup interview. It is the time before an interrupted task was picked up again, usually after about two other tasks, not the time needed to regain focus. It is often credited to a 2008 paper that does not contain the number; her peer-reviewed 2005 paper gives 25 minutes 26 seconds.

5. "Students can only concentrate for 10 to 15 minutes in a lecture." Not supported. A 2007 review in Teaching of Psychology found little evidence for it, and a 2016 review found attention varies far more with the teacher than with time.

6. "Gen Z and millennials have shorter attention spans." Unproven. Younger people use devices more heavily, but the Microsoft report usually cited for this says age is not significantly tied to sustained attention, and large studies show attention peaks in the early 40s rather than declining by generation.

7. "People only read 20% of a web page." Real but overgeneralised. It is a 2008 estimate of time available to read, from 25 users browsing in 2005, not an eye-tracking measurement of what anyone reads today.

8. "TikTok rewires your brain." Plausible but unproven. Heavier short-form use is consistently associated with poorer attention, but the evidence is correlational and long-term effects are not established.

9. "Multitasking makes you more productive." Largely false. Task switching has measurable costs, and heavy multitaskers do not get better at it.

10. "85% of Facebook videos are watched without sound." Outdated and misattributed. It came from a 2016 Digiday report quoting a few publishers' own data, never a Facebook figure. Meta now reports that over 75% of Instagram Reels views are sound on.

11. "You have 3 seconds to hook viewers." A useful rule of thumb, not a hard limit. Meta found up to 47% of campaign value in the first 3 seconds, TikTok points to the first 6 seconds, and Dentsu sees attention returns continue up to about 20 seconds.

12. "Videos over 2 minutes lose most viewers." Oversimplified. The original 2016 Wistia analysis said engagement holds steady to 2 minutes, then declines gradually, with 6 to 12 minutes a second sweet spot. Current Wistia data shows tutorials of 1 to 5 minutes still hold more than half their viewers.

What the attention span statistics mean for video

The pattern across the research is consistent. People have not lost the ability to pay attention; they have more things competing for it, and they decide quickly. In a feed, that decision happens in one to three seconds. Once someone chooses to watch, they will stay with longer content that earns it: tutorials, webinars and 6 to 12 minute videos all hold viewers well.

For anyone making video, that points to three practical rules. Put the point, or the product, in the first few seconds. Cut every section that does not earn its place, because completion rates fall as length rises. And design for sound on and captions both, since platforms differ. That is mostly editing work, and it is where most of the attention is won or lost. Our YouTube statistics show the same pattern at platform scale.

How we verified these attention span statistics

Each figure was taken from the original study, report or publisher page, not from a roundup. Peer-reviewed studies were checked against their journal or PubMed abstracts. Where an original page is offline, as with the 2015 Microsoft report and Statistic Brain's page, we used Internet Archive copies of the official documents.

A few cautions. Many attention figures come from companies that sell attention measurement, ads or devices, and we label them. Several workplace and web reading studies are small or old, but they remain the original sources for numbers that are still widely quoted. Correlation is not causation, which matters most for the short-form video research. For B2B-specific video data, see our B2B video marketing statistics.

Where Pixel8 fits

Pixel8 Production is a video editing agency for B2B brands, SaaS companies and agencies. We do not film. We cut the footage you already have for attention: hooks in the opening seconds, tight pacing, captions and platform-specific versions, which our short-form video editing service covers. Pricing is a $900 one-time trial, then $2,000 to $3,200 per month; see pricing.