What we have instead is “average commercial minute audience” ratings, which vastly overstate the actual viewership of the ads, and are not ad specific. As we have long noted, the Nielsen people meter system has no basis for determining whether a person who has identified him or herself as a program viewer when the channel is first selected is still in the room -- let alone “watching”-- later, when a commercial break appears.

Unless the people meter panel member indicates a cessation of viewing when a break begins, which almost never happens, the system assumes that all the ads were seen, on a second-by-second basis. Yet evidence from numerous observational studies tells us that this is not the case. In reality, commercial “viewing” is being overstated by 50% or more, with variations by program type, the amount of ad clutter in the break, the inherent appeal of the ads aired, etc.

Clearly this is a problem.

It is therefore incumbent on national TV buyers and sellers to use new technologies to create a parallel system. Nielsen data would continue to be used to measure TV set usage and claimed program viewing per commercial minute --more or less in the current manner -- but a second service would provide commercial-by-commercial, eyes-on-screen ratings for each telecast. Such findings would be used to “adjust” the Nielsen ratings downward for each advertiser’s buy, to reflect actual, not assumed viewing.

If a brand buys a schedule of :15s and :30s on a network in 15 to 20 different shows, and expects this to deliver about 200 18-49 GRPs, but the average ad viewing factor for these placements is 30%, then the buyer would realize that only 60 real world commercial viewing GRPs are being delivered. This is the result the seller would guarantee, not the 200 GRPs measured in the old way.

Needless to say, it will come as a shock to many brands that they have been kidding themselves for years, basing their buys on wildly inflated “commercial minute audiences.” However, once they grasp the significance of real commercial viewing data, they will adjust their thinking. Perhaps they will be motivated to spend more on national TV to get back (or at least close) to the reach and frequency levels they thought they were attaining.

There will be issues to address. For example, how is commercial viewing now defined? Is average second eyes-on-screen the best metric? We don’t think so; a certain proportion of the commercial must be seen for a viewer to get enough of a message for the brand to make a sale. The solution may be determined by correlating second-by-second screen data with commercial recall findings from the same panel members. From this research, it may be found that having eyes-on-screen for 50% or more of a commercial’s content is a more realistic metric than the average-second ratings, but this would take some work to determine.

Another issue concerns the size and composition for the eyes-on panel. Clearly, it is going to show much smaller ratings than the people meter panel, yet it is unrealistic to expect that this can be compensated for by using a panel double or triple the size of Nielsen’s. That would be prohibitively expensive. So, a compromise would be needed, perhaps settling on an eyes-on panel of 20-25,000 homes, which measures all content shown on their screens, including SVOD, OTT, etc.

Obviously, such a panel is not going to provide statistically reliable break-by-break, eyeson-screen ratings by demos for programs garnering per minute audience ratings as low as .1-.2%, and there are many of these, especially on cable. In such cases, buyers may have to be satisfied with total schedule eyes-on ratings on a gross basis, where every set usage instance is counted, but not broken out individually. Even in situations where rating size is less of a concern, the data must be used sensibly; if necessary, with aggregate findings across a skein of episodes used as a substitute for more detailed but potentially unstable data for each commercial placement. Finally, all findings must be on an absolute basis—what percentage of the audience “saw” the ad—not on a relative basis, which would only allow comparisons to the breaks when each commercial appeared.

Can such a system be developed?

There is no doubt that it can be done, not way off in the future, but right now. The question is, will advertisers and their agencies take the required initiative to really improve the relevance of their national TV buys, including funding research, to come up with an acceptable definition of true commercial exposure? Or do we just keep talking about it, but do nothing?