A YouTube Mystery, a Thumbnail Hypothesis, and a Medical Illustration Experiment
Mayo Clinic has uploaded more than 9,000 YouTube videos to its YouTube channel since its establishment in 2008. Despite new high-quality and timely videos regularly being uploaded, the video receiving the most views in a given week has lately been “New Route to the Heart - Mayo Clinic,” a video that was uploaded in 2012. YouTube metrics indicated that as of June 6, 2018, nearly 100% of its playback locations measured by watch time were on a YouTube watch page, and 68% of the approximately 1.4 million lifetime views for this video came from “Suggested videos” within YouTube. The reason for this video's great success was perplexing.
A search in February 2018 for “angioplasty” videos on the Mayo Clinic YouTube channel yielded the below results.
While not all of the videos in the search results are equal in topic covered or audience pertinence, a curious finding, nonetheless, was that “New Route to the Heart - Mayo Clinic” was the only video in the top six search results that had a medical illustration as its thumbnail.
Could “New Route to the Heart” be receiving more views due to having a medical illustration as a thumbnail? Herein, started a research project.
We identified a series of videos that had been uploaded to the Mayo Clinic YouTube channel between 2011 and 2013. All videos in this series were hosted or narrated by an individual telling a medical story, often including a physician interview and/or patient testimonial.
Forty videos were identified as meeting a variety of criteria: being publicly available on YouTube, not exceeding three minutes in duration, having received at least ten video views in the prior 100 days, not being a predominantly seasonal topic where views would fluctuate throughout the year (e.g., snow shoveling safety), not focusing on an isolated public event, and initially having a thumbnail other than a medical illustration.
Some of the forty videos did not have a medical illustration within the video, so we did not consider them eligible to be in the intervention group, which was a convenience sample of twenty videos. For the twenty intervention group videos, we changed the YouTube video thumbnails to a medical illustration pulled from somewhere in the video (see the example below of pre- and post-intervention thumbnails for a video in the intervention group).
The remaining twenty videos maintained their original, non-medical illustration thumbnail, and became part of the control group.
Two hundred days of metrics were pulled, with the pre-intervention 100-day period lasting from June 17, 2017, to September 24, 2017, and the post-intervention 100-day period lasting from September 28, 2017, to January 5, 2018. Three days in between these two 100-day periods were excluded to avoid any inadvertent viewership counts generated from reviewing and changing thumbnails, and to allow Google time to index the new thumbnails in search results.
The viewership results were as follows:
Aggregate video views from the control group increased by approximately 9.4%, and their watch time increased by about 9.3%. In contrast, aggregate video views from the intervention group increased by approximately 16.9%, and their watch time increased by about 21.1%. While we do not venture to explain the percent increases in the control group, the larger percentage increases in the intervention group seem to align with our hypothesis that using medical illustrations as thumbnails would yield greater viewership.
We conducted a statistical analysis, but with a modest sample size of forty videos with highly variable view counts and watch times, we were unable to provide conclusive evidence.
While these results do not explain the enormous success of the “New Route to the Heart” video, this study does suggest using medical illustrations as thumbnails may benefit viewership.
What approach to YouTube video thumbnail selection do you use? Have you experimented with medical illustrations as video thumbnails?