العربية Create account

Learning design · 28 October 2026 · 5 min read

Minute six: when employees leave a training video, and what brings them back

Your employees leave a training video at about minute six. That is what 6.9 million viewing sessions on edX showed. A short question after each segment halves mind-wandering.

ZERAX Academy

Share

The short version, in four numbers

  • 6 minutes: the most time the median learner spends on an instructional video, however long the video is (6.9 million edX sessions, 2014).
  • Less than half: how much most learners watch of videos longer than nine minutes.
  • 41% to 19%: the drop in mind-wandering when learners answer a question after each segment (Harvard, 2013).
  • 8 seconds: a popular attention-span figure with no study behind it (BBC, 2017).

How long should a training video be?

Under six minutes. Philip Guo, Juho Kim and Rob Rubin analysed 6.9 million viewing sessions across 862 videos in four edX courses, watched by about 128,000 learners. The median time a learner spent on a video was at most six minutes, short video or long.

The median means half of learners had left before that point. On videos longer than nine minutes, most stopped before the halfway point.

How much of a video the median learner watches, by lengthBars for five video-length buckets. The median learner watches about 100% of 0–3 and 3–6 minute videos, about 95% of 6–9, about 55% of 9–12 and about 20% of 12–40 minute videos. The share who tried the question afterwards: 56%, 48%, 43%, 41% and 31%. How much of a video the medianlearner watches, by lengthPast 9 minutes, the share watched falls below halfShare of the video watched (median)half the video100%0–3 min100%3–6 min95%6–9 min55%9–12 min20%12–40 minAttempted the question after the video56%48%43%41%31%Source: Guo, Kim & Rubin (2014), 6.9M edX sessions. Shares watched are approximate (paper's Fig. 2).
How much of a video the median learner watches, by lengthBars for five video-length buckets. The median learner watches about 100% of 0–3 and 3–6 minute videos, about 95% of 6–9, about 55% of 9–12 and about 20% of 12–40 minute videos. The share who tried the question afterwards: 56%, 48%, 43%, 41% and 31%. How much of a video the medianlearner watches, by lengthPast 9 minutes, the share watched falls below halfShare of the video watched (median)half the video0–3 min100%3–6 min100%6–9 min95%9–12 min55%12–40 min20%Attempted the question after the video0–3 min56%3–6 min48%6–9 min43%9–12 min41%12–40 min31%Source: Guo, Kim & Rubin (2014), 6.9M edX sessions. Shareswatched are approximate (paper's Fig. 2).
The longer the video, the less of it is watched, and the fewer learners try the question after it. Source: Guo, Kim & Rubin (2014).

Attempts at the practice question after the video fell with every step up in length: 56% after videos under three minutes, 31% after videos of 12 to 40 minutes.

The limits: all four courses were in maths, science and engineering, and "watching" here is time logged by the video player, not attention observed directly.

How do you keep learners focused during a video?

Ask them a question after each segment. At Harvard, students watched a 21-minute statistics lecture split into four segments. After each segment, one group answered short questions on what they had just seen, a second group read the same questions with the answers, and a third did arithmetic unrelated to the lecture. At random moments, every student was asked: is your mind wandering right now?

A question after each segment halved mind-wanderingThree groups watched the same lecture in four segments. Between segments, the first did sums and reported mind-wandering 41% of the time, the second read questions with answers (39%), and the third answered questions (19%). A question after each segment halved mind-wandering21-minute lecture in 4 segments. How often students said "my mind is wandering"No questionsMind wandered 41% of the timeSeg 1sumsSeg 2sumsSeg 3sumsSeg 4Read answersMind wandered 39% of the timeSeg 1readSeg 2readSeg 3readSeg 4AnsweredMind wandered 19% of the timeSeg 1quizSeg 2quizSeg 3quizSeg 4Source: Szpunar, Khan & Schacter (2013), PNAS, experiment 2 (48 students).
A question after each segment halved mind-wanderingThree groups watched the same lecture in four segments. Between segments, the first did sums and reported mind-wandering 41% of the time, the second read questions with answers (39%), and the third answered questions (19%). A question after each segmenthalved mind-wandering21-minute lecture in 4 segments. How often studentssaid "my mind is wandering"No questionsSeg 1sumsSeg 2sumsSeg 3sumsSeg 4Mind wandered 41% of the timeRead answersSeg 1readSeg 2readSeg 3readSeg 4Mind wandered 39% of the timeAnsweredSeg 1quizSeg 2quizSeg 3quizSeg 4Mind wandered 19% of the timeSource: Szpunar, Khan & Schacter (2013), PNAS, experiment2 (48 students).
Same segments, same breaks. The only difference is what students did between segments. Source: Szpunar, Khan & Schacter (2013), PNAS.

Students who answered questions reported mind-wandering 19% of the time, against 39% for those who read the answers and 41% for those who did arithmetic. Reading the answer was not enough. Recalling it from memory is what cut mind-wandering.

In the paper's first experiment, students who answered questions took notes on 17% of slides against 6%, and answered 84% of the final segment's questions correctly against 59%.

Is the human attention span really 8 seconds?

There is no evidence for it. The figure appeared in a 2015 Microsoft Canada report claiming human attention had fallen to eight seconds, less than a goldfish. When BBC journalist Simon Maybin traced it in 2017, the report credited it to a website called Statistic Brain, and the two sources that site listed, the US National Library of Medicine and the Associated Press, had no record of the study.

Where does the "8 seconds" claim come from?A trace: the 2015 Microsoft Canada report credited the 8-second figure to the Statistic Brain website, which listed the US National Library of Medicine and the Associated Press. Neither had a record of any study. Where does the "8 seconds" claim come from?Traced by the BBC in 20178 secondsMicrosoft Canadareport, 2015Statistic BrainwebsiteUS Nat. Library ofMedicine · APNo recordof any studySource: Simon Maybin, BBC News, 2017.
Where does the "8 seconds" claim come from?A trace: the 2015 Microsoft Canada report credited the 8-second figure to the Statistic Brain website, which listed the US National Library of Medicine and the Associated Press. Neither had a record of any study. Where does the "8 seconds"claim come from?Traced by the BBC in 20178 secondsMicrosoft Canadareport, 2015Statistic BrainwebsiteUS Nat. Library ofMedicine · APNo recordof any studySource: Simon Maybin, BBC News, 2017.
Each step points to the next, and the last step is nothing. Source: BBC News, 2017.

Gemma Briggs of the Open University told the BBC that an "average attention span" is close to meaningless, because attention depends on the task.

The other popular rule, that attention collapses 10 to 15 minutes into a lecture, is weak too. Neil Bradbury reviewed the evidence in 2016 and found that the primary data do not support it, and that attention varies more between teachers than between teaching formats.

Why is focus harder in e-learning?

Because it happens on a screen, where the next distraction is one click away. Gloria Mark at UC Irvine measures how long office workers stay on one screen before switching. In her 2004 paper it was about two and a half minutes. In studies from 2016 to 2020 it averaged 47 seconds. She says half of these interruptions are self-interruptions.

How long office workers stay on one screen before switchingTwo bars: 150 seconds on one screen in 2004, and 47 seconds in studies from 2016 to 2020. How long office workers stayon one screen before switching2004150 seconds2016–202047 secondsSource: Gloria Mark, UC Irvine (interview, 2024). Measures screen switching, not focus capacity.
How long office workers stay on one screen before switchingTwo bars: 150 seconds on one screen in 2004, and 47 seconds in studies from 2016 to 2020. How long office workers stayon one screen before switching2004150 seconds2016–202047 secondsSource: Gloria Mark, UC Irvine (interview, 2024). Measuresscreen switching, not focus capacity.
Source: Gloria Mark, UC Irvine (interview, 2024).

The number measures switching between screens at work, not a limit of the brain. It does describe where most corporate training happens: a browser tab between email and chat. A lesson that asks for an answer every few minutes gives the learner a reason to stay in it.

How do you design a lesson that holds attention?

Five rules, taken straight from the studies above:

  1. Keep each video under six minutes. Decide the cut points when you plan the lesson, before recording.
  2. Put a question after every segment. Ask learners to recall the answer themselves. Showing them the answer is not enough.
  3. Measure answers, not completion. Completion means the video played. An answer after each segment means the learner thought.
  4. Fix the explanation before the format. Attention varies more between teachers than between teaching formats.
  5. Stop quoting the goldfish number. Anyone who checks it will find nothing behind it.

Frequently asked questions

Do these findings apply to employee training?

Both studies used university students, so treat them as a strong starting point rather than a fixed rule. Measure how many learners answer the question after each segment in your own course, and adjust segment length from that.

What kind of question works between segments?

A short question that makes the learner recall what the segment just covered. In the Harvard experiment, reading questions with their answers did not reduce mind-wandering; answering them did.

Is six minutes a hard limit?

No. It is where the median viewer stopped in four science courses. Your own course data sets the right limit for it.

Are shorter segments enough on their own?

No. In the Harvard experiment all three groups watched the same short segments, and those who did arithmetic between them still reported mind-wandering 41% of the time. The question made the difference.

Sources

  1. Guo, P. J., Kim, J., & Rubin, R. (2014). How video production affects student engagement: An empirical study of MOOC videos. Proceedings of ACM Learning @ Scale, 41–50. PDF
  2. Szpunar, K. K., Khan, N. Y., & Schacter, D. L. (2013). Interpolated memory tests reduce mind wandering and improve learning of online lectures. PNAS, 110(16), 6313–6317. PMC
  3. Bradbury, N. A. (2016). Attention span during lectures: 8 seconds, 10 minutes, or more? Advances in Physiology Education, 40, 509–513. DOI
  4. Maybin, S. (2017). Busting the attention span myth. BBC News. BBC
  5. UC Newsroom (2024). Interview with Gloria Mark on attention. UC

What this research does not show: it used university students and science courses, "watching" is logged time rather than observed attention, and the 47 seconds measures screen switching, not the brain's capacity to focus.

Share

How can ZERAX help your team do this?

At ZERAX, each lesson is short pages, and every lesson ends with an activity that measures understanding, not attendance. Employees can't skip a video or jump to its end, and you decide whether activities are mandatory. Reports show completion for every department and every employee, and who is behind.

We can build this for your team. Get a proposal for your team: tell us about your organization and team size, and we'll come back with a proposal that fits. Or email us at contact@zerax.online.

Browse courses

Read next