Beware The Hype Machine: Sports Ratings VS. Esports Viewers
Headlines are exploding with esports viewership numbers. Major news outlets such as USA Today and ESPN continue to report that more people watch League of Legends on the live streaming service Twitch.tv than watch professional sports on television. A deeper dive into the numbers reveals that this excitement belies the truth behind esports viewership: esports viewership numbers are inflated and their comparison to Nielsen ratings is flawed.
As we noted previously, there are no established standards for reporting streaming viewership numbers. Concurrent viewers throughout a Twitch stream are currently being reported by the press as the best metric to measure a stream’s success. This is not a helpful approach. Here’s why:
The methods of collection differ. Nielsen ratings are calculated using statistical models derived from data sets gathered through a variety of means, including set-top boxes. Twitch utilizes web analytics which are collected from each individual that connects to a given stream.
The data is not equivalent. Nielsen ratings measure how many individuals watched the entirety of a program. Twitch’s concurrent viewers (CCU) statistic shows how many individuals are watching a stream at a given moment, whether they watched for a minute or for an hour.
Comparisons lead to inflated perceptions of esports audience size. On the surface, these statistics lead journalists to conclude that Twitch’s viewership vastly outstrips that of major sporting events. One example is the 2016 League of Legends World Championship where several news outlets reported that more people had watched the event on Twitch than had watched the World Series or NBA Finals that year. This is simply not true. Reporting like this creates a false image of the success of esports’ content relative to that of traditional sports.
Nielsen became recognized as a reliable source of audience data because it was used evenly throughout the TV industry. Comparing Nielsen ratings from two different networks produces a fair assessment because the data is derived in the same way and shares the same biases. Twitch does not publish viewership data on a regular basis, and does not claim that their analytics are in any way comparable to Nielsen ratings, but this hasn’t stopped some individuals from trying. Consensus is forming around average concurrent viewers as the most reliable statistic for how many viewers watched a stream in its entirety. Average CCU polls the number of individuals watching a stream at frequent intervals and averages them over time. Like Nielsen ratings, measuring average CCU will not reveal precisely how long viewers watched a stream, but it can be used to help understand how many viewers may have watched the entire stream. This approach, while flawed, is more helpful than the methods frequently reported in the press.
Using a more balanced analysis based on average CCU, and adjusting for a predominantly international audience, The Next Level estimates that the 2016 League of Legends World Championships viewership was closer to 7M, placing it well below the World Series and NBA Finals that year. This analysis, though helpful, is an educated guess, and illustrates why caution should be exercised when seeking to understand esports’ reach for a given stream.
When dissecting the analytics produced from streaming data, a more scientific approach could be taken. A correlative data set that incorporates not only total viewership, but also the duration it was watched, and how the stream was consumed, could produce a more accurate understanding of stream audiences. Balancing this data to include the same biases as Nielsen ratings would go a long way to making them comparable. Until this framework for comparing Twitch analytics to Nielsen ratings has been established, any claims about massive esports audiences, particularly when compared to traditional sports, should be met with a healthy dose of skepticism.
