Ratings Wars

American Demographics, Oct, 1998 by Joe Terranova

For years television programmers, advertisers, and agencies have been grousing about Nielsen Media Research, the national television ratings company. They say the sample Nielsen uses isn't representative, that detailed information doesn't reach clients fast enough, and that the service is too expensive. But despite such jawboning, the industry has been unwilling to support the steps necessary to improve the service. Every so often, a would-be Nielsen competitor has appeared on the scene, sometimes with technical innovations in tow, only to be stood up by Nielsen clients who, for all of their talk, have steadfastly refused to lend financial support to a second service.

But now the stakes are escalating. After a summer in which prime-time cable network viewers outnumbered those of the four broadcast networks combined during most weeks, it seems only a matter of time before the same can be said of the regular TV season. Naturally, the broadcast networks want to keep their share of the ad pie-some $13 billion of $44.5 billion in total TV spending- and they're hoping that a new ratings service will help. They've bet $40 million on the latest Nielsen challenger, Statistical Research, Inc. (SRI), and have promised millions more in the hope that SRI can deliver the goods, and for a cheaper price.

Four years ago, ABC, CBS, NBC, and Fox, as well as a handful of advertisers, agencies and cable networks, commissioned the Westfield, New Jersey-based SRI to create a better TV ratings system. The result was Systems for Measuring And Reporting Television, or SMART, which SRI installed in a sample of 500 households in Philadelphia. In August the networks and several advertising agencies pledged an additional $60 million of the projected $100 million that SRI says it will need to ramp up to the national level, hopefully in the next few years. But Nielsen plans to burst SRI's bubble before then with a number of its own service improvements, including a long-awaited online reporting system, DART (Direct Access Research Tools), which at press-time was scheduled to launch in mid-September.

Nielsen's national ratings service (the company also measures local TV viewing) is based on a sample of 5,000 television households, or an estimated 11,000 people. To do this, Nielsen installs two meters in each home: one for recording program data-when and where a TV/VCR is tuned-and another device called a People Meter, for recording "persons data"-who's doing the viewing. One problem with Nielsen's People Meter, some say, is that participants are required to push a button on a remote (each participant picks a number on the box by which to identify themselves) whenever they start and stop their viewing-a lot to ask of a zoned-out couch potato, a multitasking parent, or a child.

"When researchers go to bed at night thinking about issues with Nielsen, number one is the validity of that sample," says Jonathan Sims, vice president of research for the Cabletelevision Advertising Bureau. Because the Nielsen technology required to meter homes is invasive-every TV set and VCR must be opened up-many prospective candidates choose not to be monitored at all, skewing the randomness of the sample. Similarly, when a new TV set or cable box is added to a home, the household is dropped from the sample until the new equipment is wired. Not surprisingly, these households also tend to be the wealthiest or most prodigious TV viewers, valuable commodities indeed for broadcast networks hurting for better demographics and higher ratings.

Critics point out that SMART also relies on a People Meter-like box, but SRI's remote uses cute little illustrations on its remote-rather than Nielsen's numbered buttons-a more user-friendly device, they say, that encourages participation in the Philadelphia test. And SRI has adopted a more network-friendly definition of viewing: Whereas Nielsen instructs its sample participants to log in when they are watching TV, SMART asks all participants to do so when they are in the room, whether they are watching a show or not. And installing SMART technology doesn't require ripping apart every TV set and VCR in the home. Instead, SMART hardware attaches to the outside of a set and scans a new uniform television program code (UTPC), which the company developed along with broadcast network engineers.

In the past, Nielsen has also worked with customers to develop various identifying codes that are embedded in the TV video, but the UTPC would be far more pervasive and fundamental to the SMART service. Nielsen has also talked since 1993 about making better use of encoding for collecting program data, but has yet to do so.

Another advantage of the proposed SMART service may be its reporting capabilities. When SMART is up and running nationally, SRI claims it will be able to report ratings in 10-second increments as part of its basic service, affording media buyers greater insight into whether a viewer is watching their commercials or surfing around to other shows. Nielsen counters that such a level of detail proposed by SRI would be impossible to handle with SMART's current PC technology. Nielsen's DART service, for example, uses a database of 1.4 terrabytes (three years of viewing), but can only report in one-minute increments.