Binge-watching

If you’re like most people, you probably watched television more than usual the last several years as the spread of COVID-19 meant more time at home for many. Although watching TV may be a regular part of our day-to-day lives for many people, have you ever wondered why random shows keep popping up in the “suggested for you” queue? By gathering data about your current and past viewing habits, streaming services create a mountain of data, which they hope to monetize by analyzing your overall consumption.

The dramatic increase in streaming services subscriptions is responsible for much of this viewer traffic (1). Statistics professor Dr. Natalie Blades keenly observed that “this change in [consumer] access has led to massive changes in media consumption.”

Along with Blades, statistics professors Dr. Scott Grimshaw and Dr. Candace Berrett wanted to find out more about how viewers consume newly released content. Specifically, they sought to understand how releasing several episodes at once, rather than one episode per week, affects consumer watching patterns. Thanks to BYUtv, the researchers were given a rare opportunity to conduct a project that would provide answers to those very questions.

BYUtv informed Grimshaw a few years ago that they were moving toward expanding their programming from content directed to a local audience to ongoing family-friendly content directed to a national audience (2). BYUtv paired up with Adobe analytics in order to make more informed decisions about programming as they acquired content to reach a broader audience. In an effort to understand whether the exclusive digital season three release of Granite Flats (3) resulted in binge-watching, BYUtv provided the data collected from the digital-only premiere to the researchers to find out more about their viewing behavior. Grimshaw still remembers getting handed the data, especially the moment when he realized that he wasn’t aware of a way to statistically model data to determine binge-watching. That’s when he reached out to his fellow colleagues for their expertise.

Just as virologists utilize statistics to predict and track the prevalence of a disease, TV networks can predict what would make a show go viral.

Blades noted that one of the benefits of being at BYU is working in an atmosphere where disciple-scholars can come together to pursue truth. While admitting that BYU isn’t the only university where gospel living and academic study can be combined, Grimshaw is appreciative that he can share both secular and spiritual knowledge with his colleagues, as well as his students.

Working together, the team was able to differentiate between the types of binge-watching present in the study data. They could more easily quantify them thanks to Berrett, who teaches a graduate course in count data and does research with Bayesian spatial categorical data (4). Blades was then able to create a model of how a show becomes viral via binge-watching by applying her previous experience studying the spread of disease caused by bio-weapons during the 2001 anthrax outbreak (5).

Just as virologists utilize statistics to predict and track the prevalence of a disease, TV networks can predict what would make a show go viral. Applying this type of analysis to a television show may seem odd, but it is the viral nature of the content which makes the comparison an innovative and effective tool.

During the research process, Grimshaw noted that their study concentrated on two things. First, they focused on when people start watching a program after it’s released. Second, they focused on how much time passes before viewers watch the whole series. “If you give people thirty new shows to watch on the first day of the month, by the time you get to the third or fourth week nobody remembers any of those new shows and so they go unwatched,” Grimshaw observed.

By utilizing count data, the research team identified three groups of viewers: 1) those who watched the entire show in one or two sittings; 2) those who watched the show over a longer span of time; and 3) those who watched the whole show in sporadic chunks. Grimshaw, Blades, and Berrett found that Granite Flats viewing went viral with almost 43% of viewing coming within the first three days (6). The analysis of the data also led the team to identify a TV-watching behavior they term “attention cannibalism.” When two or more shows are released at the same time, for example, viewers are inherently forced to decide which show to watch. Attention cannibalism is the process of one show capturing a viewer’s attention over another show.

Photo by Jonathan Hardy/BYU Photo

The team’s research findings enabled them to determine optimal show release intervals, ones that would allow viewers to watch more new content. This type of analysis could enable streaming services to estimate when viral viewing of a show could potentially end so the service could better predict when an optimal time to introduce new content would be. Their research also provides evidence to streaming companies that releasing content in a staggered manner optimizes viewership. By keying into the role attention cannibalism can play in viewer behavior, networks can increase monthly revenue by spacing out new releases. In fact, “some new content is reverting to more traditional weekly releases instead of [a] full season release,” Blades noted.

While studies in the past have examined content from streaming service catalogs, this study focused on the release of original content. “Entertainment is no different from sports or business or anything else,” Grimshaw said regarding the role of statistics in media. “We’re all looking at how to engage people, and then at some level, trying to monetize that attention.” But data science analytics don’t just benefit streaming services’ bottom lines. They can also improve consumer experience.

As streaming services seek to understand binge-watching trends, viewers will ultimately reap the benefit as they learn how to identify optimal release intervals. After all, nobody should have to choose between binging BYUtv classics like Studio C and Granite Flats.

By Brendan Murphy

1. “Covid-19 Accelerates Digital Adoption,” transunion.com, June 2020 Media Consumption Infographic, https://content.transunion.com/v/ dgt-jun-20-infographic-media-consumption.
2. “BYUtv launches original series to bring back family-friendly programming,” news.byu.edu, BYU University Communications, March 27, 2013, https://news.byu.edu/news/byutv-launches-original-seriesbring-back-family-friendly-programming.
3. Granite Flats. Directed by Scott Swofford, Brian McNamara, and Blair Treu. Written by John Christian Plummer. BYUtv, April 7, 2013.
4. Candace Berrett and Catherine A. Calder, “Bayesian Spatial Binary Classification,” Spatial Statistics 16, ((May 2016): 72–102, https://doi. org/10.1016/j. spasta.2016.01.004.
5. Ron Brookmeyer, Natalie Blades, Martin Hugh Jones, and Donald A. Henderson, “The Statistical Analysis of Truncated Data: Application to the Sverdlovsk Anthrax Outbreak,” Biostatistics 2, no. 2 (2001): 233–247, https:// doi.org/10.1093/biostatistics/2.2.233.
6. Scott D. Grimshaw, Natalie J. Blades, and Candace Berrett, “Going Viral, Binge-Watching, and Attention Cannibalism,” The American Statistician 74, no. 4 (2020): 380–391, https://doi.org/10.1080/0003 1305.2020.1774415
7. "About BYUtv," byutv.org, accessed May 30, 2022, https://www. byutv.org/about.