Local TV Audience Measurement is Broken; How Can We Fix It?


There was a time when the sole method of TV audience measurement involved giving paper diaries to families and instructing them to write down the TV programs their family viewed. Most families diligently filled in the diary and mailed it back. It was a fairly accurate representation of what people actually watched on the three networks available at the time. This was done four to seven times a year in local markets. And, while we in the industry knew it wasn’t the perfect measurement, we knew it was probably close enough.

Diaries & Meters Try to Adapt to Changing Consumer Habits
Today 70 percent of TV viewing households are now measured by set top meters. This leaves 30 percent of the country using diaries.

Nielsen has had trouble getting accurate and stable ratings estimates from its diary sample. An overly fragmented viewing environment has led to ratings too tiny to consider accurate when compared from one survey period to another. Add to that an antiquated research methodology and reluctant respondents, and the accuracy is even more suspect. For low-rated programs, the relative standard error can be over 60 percent for certain demos. That means that a .5 rating could in reality be .2 or .8. This can have a strong impact on local TV posting records.

Technology-Enabled Options
There are several alternatives emerging, including Rentrak, a TV measurement company that relies on set top box (STB) data to estimate local viewing in all markets. Rentrak measures set viewing for all sets connected to a cable box and served by AT&T, Charter Cable and the Dish Network. Respondents do nothing — unlike Nielsen there are no buttons to push or diaries in which to write. All viewing data is sent straight from the STB, to the cable headend, to Rentrak.

The benefits of this approach are clear — it does not require respondent cooperation, and the sample size is near 20 million. Even very small cable networks report some viewing on the local level, and viewing is measured 24 hours a day, 365 days a year. At last count, more than 175 local stations had purchased Rentrak data, and several national networks had added it to their arsenal.

Modeling Missing Data
There are trade-offs, however. Under Rentrak measurement, many cable households are not monitored, including customers of Time Warner, Comcast, Cox and satellite provider DirecTV. This is mostly because some set top boxes are not capable of return path data. Rentrak attempts to model the missing data using a proprietary algorithm. A similar approach is taken with over-the-air viewing. While representing only 10 percent of total viewers, the missing household data represents 13 percent of broadcast network viewing according to Nielsen Media Research.

Also worth mentioning is that STB ratings measure viewing by set and not demographic viewing. Rentrak models viewing based on household demographics provided by a third party. A household with a teenage boy and a mom age 45 would yield viewer data for both parties, regardless of the programming.  As a result, ratings are up to three times higher than Nielsen data. Local stations obviously appreciate this boost!

Most, if not all, national players are continuing to use Nielsen, even if they complement it with Rentrack data. Only a few small stations have dropped Nielsen.

No Silver Bullet for Local TV Audience Measurement
Regardless of Rentrak’s drawbacks, it is clear there is a growing dissatisfaction among TV buyers and sellers, and a belief that Nielsen is not providing accurate audience numbers, particularly in local markets. Nielsen recently announced a plan to improve its local ratings through increased sample size, more passive monitoring and utilizing STB data in a limited way. These changes are due to be rolled out beginning in late 2013.

And while Nielsen has recently announced an alliance with Twitter to report social TV ratings, this new product will measure social conversation around TV programs, not the audience viewing them.  Similarly, Nielsen’s proposed merger with Arbitron, which also relies heavily on diaries for radio listening measurement, is unlikely to provide more accurate local TV ratings.

Despite the desire for a perfect data source for local TV ratings, the industry must work with the limited options available. Marketers should understand the limitations of the data and look forward to improved measurement in 2013.

Want more content like this? Subscribe now!

Julie Pahutski
Julie Pahutski

Julie Pahutski leads Empower’s Decision Sciences practice, which is responsible for media measurement, market research, and digital/web analytics. Her passion is data, whether it is for advanced targeting, media analysis, attribution modeling, or dashboarding. She spearheaded the creation of many proprietary tools at Empower, including Impact Moments® and MAPS (a geographical targeting system).