Well, maybe so. The problem is that every year this amazing game is: (a) built on a steadily declining base, and (b) based on less and less solid ground.
Prime time as a daypart has already declined precipitously since everybody gathered around their sets and watched "I Love Lucy" together in the 1950s. By 1975, prime time represented only one third of all household tune-in time. Last season, it was down to 24%. Its share of advertising dollars is, of course, considerably higher than this.
Then we have the big broadcast networks. Another new record was forged in 2001-02, when prime-time household viewing shares for the "Big Four" fell below 50%. They have plummeted to 46% from 72% 10 years ago.
Combine these two facts and all the headlines, the gossip and the fuss over the upfront turns out to represent about 11% of what TV viewers actually watch.
OK. So we know all this. We know audiences are fragmenting; that digital cable and satellite TV are now in one third of homes (making the problem worse); and that personal video recorders are slowly creeping in, waiting for their tipping point.
A straight-line projection might even suggest that the annual upfront-which is mostly about grabbing a handful of the remaining 20 to 30 shows with a worthwhile rating-will soon be down to haggling over 10 to 15 shows, and maybe 5% of viewing time.
looking under rocks
Does anyone care to look underneath some of the other rocks in our path? The metrics we all rely on to tell us this dismal story of declining broadcast network ratings and shares are themselves becoming more suspect every year.
Nielsen Media Research generates national rating numbers for around 2,200 programs every week. Between 20 and 30 of these shows capture 5% or more of the adult viewing population (mostly in the broadcast prime hours). But 90% of the shows get less than a single rating point.
With a sample base of around 4,900 18-to-49 year olds on its national people-meter panel, this means that fewer than 50 panel members are actually watching most programs. And once we get into the nonsense of analyzing programs and networks to two decimal places-so we can get a number to talk about-we are down to just a handful of people.
If the size of the sample is a problem, so is the quality. Of every 100 or so people approached to take part in the Nielsen panel, somewhere between 30 and 40 agree to participate. These participation rates are under severe pressure and continue to undermine the accuracy of all media research. Nobody really knows whether the people who refuse to take part are different in any way in their viewing habits than those who agree to participate. But nobody is complaining because Nielsen panels are probably biased toward people who are more interested in watching TV. That suits the people paying Nielsen's bills.
Secondly, the fragmentation already referred to means that 90% of the TV program ratings generated are at best problematic (whatever statistics theory may tell you).
Finally, what will be the point of prime time anyway when PVRs do take off, in the next two to three years, and people start recording and watching programs at different times? Prime time will be a different experience for everybody. People might want to watch "ER" on Saturday mornings and "West Wing" late at night on Mondays. And, of course, they may not care to watch the ads.
a potential solution
Given the way PVRs "search" for programs to watch, broadcast TV will no longer have any advantages whatsoever over cable. Navigation will be easy: A machine will do it while viewers are at work or doing something else.
There is a potential solution to the measurement problem (if not the behavior) here, although precious little attention is being paid to it. Eighty-five percent of American households are equipped with set top boxes of one sort or another. It is at least theoretically possible that the problems of sample-based research, outlined above, could be partially addressed through some kind of actual counting of the number of household sets tuned into any program on any station in any place. The counts could be minute-by-minute or even second by second, giving solid data on which sets were tuned to which ads when.
There are many hurdles to jump before this could become a reality. There are several incompatible types of set-top box, many of them quite old. There is the problem of measuring non set-top box homes, of aggregating and processing the information and of privacy concerns.
Most important is the question of measuring individuals within the households. This would inevitably still have to be a sample-based method, combined perhaps with viewer modeling of one sort or another. But at least it would be based on solid ground. An analogy to this is the print medium, where audited circulation data is complemented with sample-based readership research.
If the above were done, we would no longer need the Nielsen Station Index in its current form, and we certainly wouldn't need a separate Hispanic service. All could be combined and integrated with the national people meter service under its current or a future form.
As we've seen in Boston, and may see again in Philadelphia, all the usual suspects will turn out to oppose such a change because it threatens the artificial dominance they have long enjoyed. Technical reasons will be given. People will yawn, and they'll turn back to their daily work, doing it the way they have always done it.
Let's get together and do something about it.
Andrew Green is managing partner, director of communication insights, for Omnicom Group's OMD, New York (email@example.com).