Have you ever wondered how viewers respond to a particular movie or commercial? Does it retain an audience or do people leave straight away? Wouldn’t it be useful to know specifically what parts of a movie or commercial are popular so that your next video posting can be better than your previous?
Within my own company, Applied Brain and Vision Sciences, we use StepForth’s web marketing savvy and tools like YouTube and Google Analytics to help us direct our own marketing approaches and help us identify market opportunities for technology development. One of the distinct advantages of having StepForth house my own company is our physical proximity to each other; there is a wonderful on-going cross-pollination of ideas and know-how.
The data represented in Figure 1, to the left of this paragraph, corresponds to viewing behavior. Overall, the data show that the video is largely able to retain an audience, possibly because the opening question posed 5 seconds into the video, “What happens when you open a can of Coke do at 14,000 feet?” The data show that viewer retention stays above average until we reveal, just before the end of the video, what happens when we open a can of Coke at 14,000 feet. Click on the video your self to see how events in the video correspond to events in the plot. Figure 1. View Retainment Plot for, “Humans vs. a can of Coke”.
According to a description on the YouTube site, the plots provided by this type of analysis show “the ups-and-downs of viewership at each moment in your video, compared to videos of similar length. The higher the graph, the hotter [the] video: fewer viewers are leaving your video and they may also be rewinding to watch that point in the video again. Audience attention is an overall measure of [the] video’s ability to retain its audience”.
Now that we have some data, what do we do with it? After collecting data, they must be interpreted.
Interpreting The Data
It isn’t enough to simply have data represented in a figure. The data must be interpreted. Data interpretation is perhaps the most important step because it the interpretation is what will generate conclusions and inferences of why the data look the way they do. The interpretation will also inform any hypotheses that are made and will guide any future work based on these hypotheses. For example, looking at this promotional video, the first question we asked our selves was, “How come the plotted line drops of so dramatically towards the end of the video when the “credit and link to our adventure blog appears?” The answer might be that viewers don’t have any interest in viewing the link to our adventure blog at the end of the video and since the video is essentially “over”, they stop watching.
We also see from the data that viewer retention on average diminishes throughout the duration of the video. While this is unfortunate, we are still retaining viewers at a level “above the average” indicator. Because of the loss of viewers, it was probably a good decision in our design to include a link to our adventure blog at the beginning of the video.
There is also a peak at the beginning of the view retention plot relating to the beginning of the video that might be separate from a regular and gradual abandonment of viewers. It is possible that some amount of the peak is due to viewers stopping the video in the middle and re-winding to the start of the video to see “who made the video and where to get more information” — our adventure blog. This of course is an optimistic view.
An alternate consideration of why we have a large peak at the start of the video is the fact that the audio quality at the start of the video is terrible. This is unfortunate but was a constraint we were stuck with. As a result of the low audio level at the beginning of the video, people could be “rewinding” the playback of the video a couple times to help them understand what is being said. (To improve upon this situation, we have added some sub-titles to the video.)
Comparing Videos
(1) a television interview video where we are explaining some of our past spatial cognition research. This video has been somewhat popular and is fairly informative. What is noteworthy is that according to these data, viewers might be skipping ahead of the TV channel introduction of the interview to get to the good stuff in the video. Perhaps a re-launch of this video, where we go directly to the “good-stuff”, might not have the below average rating at the start.
Figure 2. View Retainment Plot for, “Television interview”.
(2) a very simple video illustrating a rotating model of brain function related to some spatial navigation research. According to the corresponding viewer retainment plot, this might be the worst video we ever posted.
Figure 3. View Retainment Plot for, “Example results of MOST-EEG”.
If you have any questions of comments about anything posted in this article, feel free to leave a comment here and add your email address if you want a more direct reply.