| ||||||||||||
PRINT
COMMENT
SUBSCRIBE
RSSThe plea followed the presentation of an analysis by Foster showing that smaller cable networks must reach disproportionately more of Nielsen's panel households to generate comparable ratings. Foster implied this was due to a methodological glitch, and the fact that Nielsen's total national sample of about 14,000 households is not nearly large enough to accurately represent smaller TV outlets as the medium continues to fragment.
"The drops in viewing can be precipitous," Foster said of the impact on smaller channels, and noted that it would become even more exacerbated in February 2009, when the federally mandated transition to digital broadcast spectrum occurs and many local broadcasters begin multicasting as many as six niche channels in their markets, fragmenting the ratings pie exponentially and putting an acute strain on Nielsen's sample households.
Current's Falcon offered a glimpse of that future yet to come, revealing her own frustration after subscribing to Nielsen's preliminary ratings for Current TV over the past couple of years, which she implied still is far from ready for prime-time.
"We noticed a lot of challenges with the numbers," she said. "There were crazy swings in household and demographic numbers in regards to concentration - really high highs and really low lows month-to-month."
She implied the dramatic fluctuations were due not to actual shifts in audience viewing patterns, but to instability in Nielsen's ratings methodology and the tiny base it uses to project national TV audiences.
"How does a network account for some of the things we came across," Falcon implored, adding that the "accuracy and usability" of Nielsen's ratings are, "questionable for us at this time."
She said Current TV is turning to some new players for a solution, especially TNS, the research company that was recently acquired by Madison Avenue giant WPP Group. TNS has been developing a new TV ratings system derived from actual digital set-top data from cable and satellite TV operators. It's big payoff, she said, would be the launch of TNS' Direct View service in January, a ratings system that will launch with an initial base of 100,000 DirecTV households in its sample, and which plans to expand to 200,000 soon.
She said Current TV also is exploring other options, but that each of them as certain drawbacks, including TiVo's audience measurement service, which is based exclusively on DVR households, and Rentrak's audience service, which grew out of a tracking system for the TV industry's pay-per-view and video-on-demand sales.
In a lively, spontaneous debate among representatives of TNS, TiVo and Rentrak, all three said they were moving rapidly to fill the void and to address their own respective methodological concerns.
Asked by MPG Collaborative Alliance host Mitch Oscar if anyone from Nielsen had anything to add, the room was silent.
RSS FEEDS
There is a 'delicious irony' in such scenarios. The smaller the channel, the more sample you need, but small channels simply do not have the money to pay for the sample! The second irony is that small channels tend to have less regular viewers which means that the variations day-to-day (that genuiney exist) are exaggerated.
Some simple maths gives a rough idea of the impact. If Nielsen have 14,000 households, each with an average of 2.5 people in the household we have a sample of around 35,000 people (feel free to correct my data as I am in Australia and don't have the actual statistics). These 35,000 people represent the US population of around 305,000,000 people. That is each person in the sample represents around 9,000 people - that is, this is the 'average weight' of a person. You only need half a dozen people to not watch (say ... go out for dinner or to the game) and that's 50,000 gone off your rating! 50,000 when you're on the bug broadcasters is nothing, but to a small channel it is a disaster.
So what is the solution?
The obvious answer is the unaffordable one otherwise it would have been done ages ago - buy more sample. Those that NEED more sample (and everyone does!) are the least able to pay for it.
My belief is that we need a "hybrid" system that takes return path household tuning data (not people viewing data from a panel) and use that data (once it has been 'cleaned' to remove tuners that are left on 24/7 like mine is) upon which to base the PEOPLE projections.
The samples are creaking under the weight of all the channels, but the tuning data of each channel by itself is not "rich" enough (or 'people-centric' enough) to paint the whole picture. Clearly we need a combination of both going forwards.
John Grono GAP Research Sydney Australia