Commentary
People Watch More TV Than Ever -- But Is It All Reported?
- by Steve Sternberg , Featured Columnist, December 7, 2016
Back then, even when there were far fewer choices — no original scripted cable series, no Netflix, Hulu, or Amazon Prime, and no regular series on premium cable — people watched only slightly more than two out of four episodes per month of the most popular series. There were no DVRs on which to store programs, or On-Demand to catch up on episodes. If you missed an episode of your favorite show, you had to wait until summer repeats.
When DVRs started to become a thing, I wrote an article speculating that they would cause television usage to go up as people started to watch that third or fourth episode per month of their favorite shows.
advertisement
advertisement
That has, to a large degree, been the case. The advent of on-demand programming has also had an impact on increasing viewership (albeit a smaller one).
Once Nielsen started including DVR homes in its sample, roughly 95% of all DVR playback was done within seven days of the original broadcast. This is why one of the main debates was between C3 and C7. With quick analysis of program performance and television buys being desirable, that extra 5% could be ignored.
Today, there are a lot more people who binge-watch several episodes of a weekly series. This means, of course, that there is probably a tremendous amount of television viewing that is not being counted in Nielsen’s currency measurement (even if C7 is used instead of C3).
I conducted a survey of my Facebook friends (a surprisingly diverse group), just as a fun exercise. I asked four simple questions about their DVR, on-demand, and OTT usage. Twenty-five percent of respondents did not have a DVR (the national average is about 50%). Here are the results.
On average, they currently have eight programs queued on their DVR (ranging from three to 20). There are currently four shows on average where they have two or more episodes stored (ranging from one to 11). Two-thirds of respondents watched at least one program on demand in the past week. Two-thirds also watched Netflix, Hulu, or Amazon Prime in the past week, and spent 5 ½ hours doing so (ranging from one to 10 hours). This seems like a lot of viewing not picked up and included in currency measurement. If you have eight programs on your DVR, you are likely not watching all of them within three days of their initial broadcast. If you have four weekly programs with two or more episodes stored, by definition you are watching many of them more than a week after their initial broadcast.
Of course this is far from a scientific survey. I’ve always believed, however, that what people are doing, what they are watching, how much they are watching, and what devices they own are correlated with, in addition to age and sex, such things as geography, income, education, presence of kids, etc. But once they have the device, and once they are watching a program or video on that device, how they watch TV is mostly contingent on age/sex.
In other words, there is no reason to think a 25-year-old in Wisconsin, who is watching “The Walking Dead” on her DVR, is fast-forwarding through commercials at a different pace from her counterpart in New York (if anyone has research that shows otherwise, I’d love to see it). So a sample that represents the country at large is much more important to measure some things than it is to measure other things.
If nothing else, this should be a conversation starter.
Dude, you forgot about VCRs. People had em and used em. And still they only watched two episodes a month.
But even then, Nielsen couldn't measure VCR playback.
Steve, correct me if I'm wrong as I believe that you were involved in this re VCRs. As I recall, it was decided to count VCR taping as equal to viewing, though how they determined who was "watching" is a pulllement. Yet, didn't Nielsen also report that a very high proportion of VCR taped episodes were never viewed?
As for long delayed binge viewing, we should remember that most delayed viewing is for primetime broadcast network shows, some cable programs and various specials. All told, these may account for 25% of a typical adult's total time spent watching commercial TV. As for the rest--early AM fare, daytime stuff, game shows, news, sports, etc, DVR activity is relatively slight. So the question arises, in terms of total volume of viewing time, how much might Nielsen be missing by not going beyond one week, not just for primetime but for all TV? As a guess, I'd say that we're probably talking about 2-3%. OK, so maybe it's higher for prime---say, 5%. I wonder if Nielsen has a report on this? Anyone know?
Make that puzzlement not "pulllement" in the second sentence of my reply. Sigh!
Ed, with VCRs Nielsen was only able to measure household recording and ascribed that to all demos. Thay were never able to measure playback, and I never saw good research that showed how much was taped on VCR was actually played back.
Nielsen does measure 35 days of playback now, but you have to pay extra to have access to the data. I believe for some heavily recorded primetime series show as much as 10% or more playback after 7 days.
There is a huge gap between what's available in Nielsen or Rentrak data (independently collected and reported for the purposes of monetizing advertising by broadcasters, agencies and brands) and the consumer consumption and usage data that has been closely held by the more technically savvy distributers (data that can include all manner of consumer usage and behavior across DVR, set-top box, SmartTV, game console, and/or mobile devices).
Netflix, Tivo, Amazon, AT&T, Roku, TimeWarner, Apple, etc can analyze very different data than the rest of the ecosystem does via third-party reporting. Decisions made the past few days, weeks and months more than illustrate this point. Add third party data and addressable buying technology to proprietary census data sets, and the only question should be: how much longer until the game has enough scale to really shake things up?
Edward, as you may be aware, I agree with you regarding the lack of scale---commercial GRPs--that hinders the expansion of "addressable TV". But we must also recognize that the use of set usage, not viewing data is another great limitation. It doesn't matter if your "big data" set-top-box ratings are based on samples hundreds of times larger than Nielsen's ratings if you don't know who is watching.As a result, except in rare cases where the advertiser can provide email addresses of its customer base, the solution is profiling---based on geography, income, ethnicity and other factors, not person by person ad exposure data.Until that problem is dealt with as well as obtaining more GRPs to sell, there will be only very slow movement.
Ed - yes we agree on quite a few things.
My point was that the digital distribution leaders (Amazon, AT&T, Netflix, Roku, TimeWarner, Tivo, etc) do know - they already store, analyze and connect to the household, and can offer deliverable audience pools across screens with better compiled HH brand and store transaction data (via Acxiom, Facebook, Neustar, Walmart, etc) than what most brands have attempted themselves. Triggering ads will be HH behavior and brand based, not program or content or demo based.
Who needs ratings when you have audiences?
Steve and Ed, just as an FYI, we handled the Nielsen meters in the old VCR slightky differently. We were able to meter the record head but not the playback head. That is (i) we knew that someone in the home was interested enough in a programme to go to the trouble of recording it, but (ii) we didn't know who in the home it was, and (iii) we didn't know if/when they palyed it back as the source material wasn't time-stamped.
So we took the simple route of crediting just the HH rating, and nothing was credited to any people/demographic rating - advertisers just wouldn't wear that.
Edward, you seem to have confused 'audience' with 'audience pools' (your term). An 'audience pool' is also called a household. As Ed explains correctly, that is a HH tuning rating which is a blunter instrument than an audience viewing rating. You may be able to attribute 'usergraphics' to a HH (such as buy a certain cleaning product brand, high usage NFL household etc.) but it is a viewing attribution model at best.