Testing times for audience measurement
December 22, 2015
By Larry Gerbrandt
Digital has been a decidedly mixed blessing for the television industry. Digital compression has led to the creation of hundreds of new channels, HDTV, IPTV, streaming media (from YouTube to Netflix), video-on-demand and digital video recorders. It has also made television viewing measurement a nightmare.
While much of the recent growth in the industry has been driven by subscription-based services, television programme creation is still heavily dependent on advertising, which in turn is based on usage measurement—ratings—and equally important—the demographics of the viewer.
In theory, digital video should provide a quantum increase in the accuracy of media measurement. Set-top boxes and DVRs can be remotely polled, providing second by second consumption. Web-based video consumption can be tracked through server logs. So why are media executives and research departments so frustrated by a medium awash in terabytes of viewing data?
The first problem is that an increasing amount of video is no longer being watched live, either though VoD, a DVR or web-based streaming media. Advertisers are reluctant to pay for time-shifted viewing, though there is a case to be made that schemes that allow viewers to fast forward through programme content but not commercials can make this limited ad inventory more valuable, especially when more consumers are deploying ad-blocking technology on their computers and smartphones.
The second challenge is that the proliferation of viewing choices has dramatically fragmented live viewing. With the exception of live sports, virtually every television programme in prime time on a major US broadcast network has suffered a loss of viewers in the latest TV season. ‘Binge viewing’ has become commonplace. Viewers are willing to wait until a TV season is over so they can watch all the episodes in a single sitting, either stored on a DVR or downloaded from a site like Hulu or Netflix, and the latter doesn’t carry advertising.
The biggest challenge, however, is that the vast majority of viewing data being collected is flawed or lacking criteria that most advertisers deem essential. Data collected from set-top boxes often can only report the channel the box is tuned to, not whether the TV set to which it is connected is on or off. Researchers have attempted to overcome this limitation by dropping boxes from the sample when the box doesn’t change channels after a specific amount of time (typically four hours) but this may undercount certain channels such as those that cater to business news and children’s programming. The biggest flaw is that neither server logs or set top box data can measure how many individuals may be watching a screen and their demographics…and advertisers rarely buy households, they buy specific demographics.
Throughout most of television’s history, the viewing data collected by companies such as Nielsen from a national sample of households has been adequate for the largest networks or the leading broadcasters in the largest markets. The splintering of viewing over the last decade, driven by the proliferation of digital technologies, has been steadily eroding the accuracy of the data. Part of the solution is to expand the size of the sample and to include a wider array of viewing devices, but that comes at considerable expense at a time when advertising revenues are under pressure (a recent survey by Advertising Age of 61 returning US prime time series showed 37 had suffered declines of as much as 51% in 30-second unit prices).
A major initiative is underway in the US, led by Nielsen, to track viewing of television series across devices for up to thirty days after being premiered, as long as the shows are digitally tagged with the appropriate programme metadata. The new reporting will also include the demographic data on which most media buys are based.
This may fill in some of the growing holes in viewing of the most popular TV series on the largest networks but the bigger challenge is the hundreds of SVoD channels springing up on web and viewable through web-connected TV sets. Though the most popular of those channels are subscription-based (such as Netflix), the bigger challenge lies ahead: measuring and monetizing micro-audiences that cumulatively are becoming a statistically significant force in the future of the medium.