Are They Really Watching? A Study of Student Video Viewing Habits

Concurrent Session 3

Session Materials

Brief Abstract

Conventional wisdom says that the shorter a video is, the more students watch. A two-year study of student video viewing habits is presented here, attempting to find the “sweet spot” in length for lecture content. The study also explores the effectiveness of segmenting a long video into shorter sections.


Josh Lund is an Instructional Technology Consultant at DePaul and a former teacher turned mad scientist. After completing a B.M. in Music Theory/Composition at St. Olaf College and an M.M. in Composition at Northern Illinois University, he spent six years teaching instrumental music at Elgin Academy, William Penn University, and Central College and worked as an active performer and clinician before returning to Illinois to complete a second master’s degree in instructional technology at Northern Illinois. A life straddling two different disciplines, technology and the fine arts, has led him to researching teaching technology in the collaborative arts, multimedia and recording technologies, and user interface design . He is really enjoying the fact that his job lets him play with technology tools all day and then teach others to use them. Josh still writes and performs on occasion, teaches the occasional wayward bass or guitar student, and is an avid gardener and disc golfer. He enjoys cooking, travel, and the outdoors, particularly when his family is also involved.

Extended Abstract

Description and Goals:

The single greatest complaint we get from instructors about their videos for courses is that students are not watching them, or at least are not watching enough of them. They begin to question the value of this type of instruction, because they are not seeing any benefit to putting all their lectures into a virtual form.  To some, this seems to reinforce their anti-online bias, when in reality, there are numerous strategies that could be employed to increase viewership and information transfer.

This study analyzes the data from our University’s Panopto server, focusing on viewership statistics.  Data analysis has been performed to isolate videos that have the highest overall viewing percentage versus the longest amount of time.  In particular, I am searching for a “sweet spot,” at which the viewing percentage will be the highest for the longest possible video.  I also am analyzing the results of chunking, or segmenting a long video into shorter segments, with the hope that segmenting a video will lead to higher viewership of the material overall.  The goal of this study is to find best practices for creating better online video content, that will be better-viewed by students.  If you make a video for your students, I would like to be able to tell you just how long it should be if you really want to keep their attention all the way to the end.


This presentation is appropriate for faculty, instructional designers, and administrators who have an interest in best practices for online video content.  There are stakeholders in each of these roles concerned with delivering video that is efficient as well as informative, and the ideal length for a video is an oft-debated question among faculty, students and instructional designers. The results of this study should be considered fairly typical for any large University, and so there is immediate applicability for the findings.


I sought to answer two main questions:

  1. What is the ideal length for a lecture video in order to ensure maximum viewing by students (both in terms of overall views and in complete viewings)?
  2. Does segmenting a long video into shorter segments improve the overall viewership of the video content?



Data was pulled from reports available within our Panopto server, including viewership and creation data on individual videos, video “folders,” and other media types, including audio only content.  All data was visualized, first comparing the video’s length with its overall viewing percentage (average minutes viewed/video length).  From here, it was possible to isolate the video content that had 75% or better playthrough, and look at the band with the highest concentration of viewers and highest playthrough.  The trend line would show the final “sweet spot” figure.

The same analysis was conducted with folders containing anywhere from one to eight or more videos.  All videos and folders are given an alphanumeric code, so there is no way to differentiate by discipline, instructor or course.


  1. The “sweet spot” is calculated to be 7’52”.  Interestingly enough, this is almost exactly how long we go between commercial breaks on prime time TV, and is the length of a Pecha Kucha talk.  The “ideal” length of a video, accounting for some differences in mean/median, is between about 3 and 10 minutes.
  2. Segmenting does indeed have a positive effect on viewership.  In fact, it increases as the number of videos increases, up to about 7 videos, at which point data is less conclusive.


This study provides tangible proof to the statement, “A shorter video will be better viewed.”  Although we have videos on our server that range from a few seconds to 4027 minutes in length, almost nothing longer than 60 minutes is ever watched in its entirety, and to be honest, no one reliably watches anything over about 20 minutes.  It is not a surprise to me that the “sweet spot” is that short.


This was a good “Phase I” for a study.  I looked at data just for the numbers, but did not take into context the subject matter, the method of delivering the material, or any enhancements made to the video.  While viewership data was mostly consistent for length vs. viewing percentage, there are always exceptions at every level, which leads me to suspect that there are also qualitative reasons why some videos are being watched more, or at least more fully, than others.

Phase II of this study will begin to examine the following factors:

  1. Discipline (which Department’s classes are these? Are they more specialized interest classes or primarily background courses such as “History of…”, or “… Literature,” etc.)
  2. Instructor style (dynamic, on camera presence, special tools, vs. low presence, slides only, etc.)
  3. Content (lecture? Tutorial? Film/documentary clip? New vs. review?)
  4. Some videos have more than 100% viewership, indicating that they were watched more than once (at least one 100% viewing, and more after that).  Why?  What about these videos made them worth watching repeatedly?
  5. Data is also available for faculty and student video creation, and I would like to do a trend analysis to determine when faculty and students produce the most content, to better leverage campus resources to meet demand.