Clean Up The Sweeps

by Dave Zornow

The sweeps, Nielsen’s February, May, July and November surveys of all DMA markets, were designed to provide media buyers with representative TV ratings in all local markets. But over time they have been turned into an atypical, stunt- and star-studded spectacle by broadcasters, networks and syndicators which inflate prime time program ratings by as much as fifty percent. And since demographics are only measured during the months when broadcasters break out their biggest programming and promotional guns, local cable ratings suffer.

By now, just about every TV viewer in American knows when a sweep is taking place. Regularly scheduled prime time programs don’t run as scheduled. Local newscasts are filled with more-than-usually captivating stories and "news" features which coincidentally relate to their movie lead-ins. During the rest of the year, viewers know they have less than a fifty-fifty chance of seeing a new episode, but during February, May and November the re-runs are few and far between. "There is no question that the networks stockpile their best programs for the sweeps," says Steve Sternberg, vice president of broadcast research and a senior partner at Bozell. "Over the last decade there has been an increasing number of repeats in non-sweeps months," he says. Sternberg adds that new programming is only part of the problem. Many stations run contests during November, February and May which are also designed to increase viewer tune-in and distort the ratings.

To their credit – and increased profitability, too – Nielsen has substantially increased the number of markets which use electronic household meters to report television viewing 365 days a year. When Nashville goes online next July, Nielsen will be using household meters to measure local market TV viewing in 59 percent of all U.S. TV households. But household meters are only half of the ratings equation. Nielsen still uses paper diaries issued during the sweep periods to collect local market demographic viewing.

How much do the sweeps distort a media buyer’s picture of TV viewing? During last season, the average prime time program rating during the three main sweeps and the beginning of the new season was as much as 52 percent higher than the rest of the year. Seinfeld’s average TV HH rating during February, May and September through November was 22 percent higher than the rest of the year (21.5 compared to 17.5). Friends was 25 percent higher (18.6 versus 14.8) and Home Improvement delivered 19 percent more households (16.3, 13.7). Rating hype also affected dramas (E.R. recorded 32 percent higher ratings), news programs (20/20: 13 percent, 60 minutes: 14 percent) and younger-skewing FOX shows (90210: 49 percent). Lois and Clark proved to be the superman of the new season and sweeps with a 52 percent higher rating than the rest of the year.

The impact of new episodes, programming stunts and specials is best demonstrated in the local markets. In Los Angeles, the combined prime time for the ABC, CBS, NBC and FOX stations during the 12 months ending last June was 30 percent higher during the February, May and November sweeps and the beginning of the new season than the rest of the year. Among cable households in Adlink’s LA interconnect coverage area, the four affiliates delivered a combined 34.5 rating compared to a 26.4 for the rest of the year. Los Angeles independent stations also showed small gains during the three sweeps and the new season months posting a four percent higher rating than the rest of the year. Cable and unreported stations, however, took it on the chin: the aggregate rating for cable and all other stations was 17 percent lower during these months. Cable/Other delivered a 20.4 rating (37 share) during the non-hyped months but only a 16.9 / 28 share during the heavy broadcast promotion months.

Ratings distortion during the sweeps is more than just a cable problem. "The relationship between network affiliates and indies is likely to be different in sweeps and non-sweeps," says Stu Gray, senior vice president of media research services at BBDO. The Adlink data back up Gray’s hypothesis: affiliate and indy share of viewing splits 57% / 15% during the five high-promotion months but changes to 47% / 16% during the seven mostly re-run months. It’s also a problem for media buyers who have to assume that demographic viewing to affiliate, independent and cable sources is the same across sweep and non-sweep months because Nielsen doesn’t provide any demographic data outside of the regular survey periods.

Is there a remedy to the broadcast-biased sweeps survey? "Continuous measurement is the only way to address this problem," says Rusty Taragan, executive vice president at Nielsen Media Research. "We do think it is something we should be talking to our clients about." Clients, like Michael Alvear at Charter Medical in Atlanta agree. "A continuous three-month survey makes a lot of sense. It will flatten out the ratings you get, providing numbers that are more steady and stable."

There is a precedent in the ratings business for moving from one-month books to continuous measurement. To reduce the effects of radio station contests and promotions, Arbitron Radio dropped one-month studies in favor of four 12-week surveys in the early 80’s. Agency researchers concede that continuous TV measurement is a good idea but they are not optimistic about its prospects. "Ideally we would like to see more measurement throughout the year, but Nielsen isn’t a non-profit organization," says Bozell’s Sternberg. Whitey Chapin, vice president for broadcast research at TN Media, agrees. "Will the local stations pay for more measurement? Are they going to pay more for a new system that might show a lower number? I don’t think so." Chapin says that although it sounds like a research issue, it simply a matter of economics. "Nielsen is not going to do this on their own to be nice. It all boils down to money. The system will only change when the stations feel the system should change."

Questions? Comments? Please write to me at diz@bellatlantic.net.