Throughout the history of television audience measurement, most people had access to the same viewing platforms and devices. When something new came along to either add additional channels or enhance the TV viewing experience, most people eventually bought it. This remained true through 2000, as VCRs reached about 90% of TV homes and DVRs were new technology. Measuring who was using each medium and device (although not necessarily how they were using them) was relatively simple.
Today’s media world is significantly more splintered. Everyone doesn’t get everything anymore.
- DVRs have been around for over 20 years, but nearly half the country still does not own one.
- People with internet-connected multimedia devices are using them more and more, but these devices are currently in only about 40% of homes (as are enabled smart TVs).
- Roughly two-thirds of TV homes have Subscription Video on Demand (compared to just 40% three years ago), and it will likely grow further as new streaming services from Disney, Apple, Warner Media, and others become available this year and next – but less than half the country will have more than one streaming service (and probably only Netflix will be in more than half of U.S. TV homes next year).
In a few years, a third of all viewers might be heavy mobile TV viewers (mostly 30-year-olds and younger), while the rest of the country may watch TV that way only occasionally. Viewing TV content online will be heavily done by a certain segment of the population, but the growing 45+ population might still be watching significantly more TV on a television set. What, how, when, and on which platform people watch video will vary by content provider, platform, device, income, life stage, and numerous other factors. Age/sex will still be important, but will no longer be the sole driving force (as it was pre-2000).
Current Audience Measurement Not Nimble Enough to Handle Rapid Change
Until the early 2000s, the pace of change was relatively slow, enabling cumbersome media measurement conglomerates to trudge along with serviceable audience measurement. As cohesive groups of people aged, their media habits were largely predictable.
It’s hard to imagine now, but before people meters debuted in 1987, the national Nielsen television sample was only 1,200 homes, and demographic data was only available 36 weeks of the year. In a three-network, one-screen, single platform, 15-channel, live-viewing world, this was acceptable.
Here’s a brief look at how the video media landscape has evolved since then:
- During the late 1980s, VCRs started to become prevalent, eventually being owned by 90% of U.S. TV households. Watching pre-recorded videocassettes became the first major new use of the television set since videogames. This contributed to the broadcast networks giving up on original scripted series on Saturday nights (which was now the biggest movie rental night). They subsequently cut back on airing made-for-tv and theatrical movies as well. Time-shifted viewing, now possible for the first time, remained relatively minor (about 5% of overall primetime viewing).
- The home VCR became the fastest growing electronic device since television itself, and caused real problems for audience measurement. It marked the first time Nielsen had to admit it couldn’t measure an element of television viewing. It was able to capture VCR recording, but playback among individual people remained beyond its scope.
- In 1990, the average home could still only receive 33 channels. Viewing habits remained relatively stable, and aside from VCR usage, there weren’t many major challenges to audience measurement. As cable and satellite television expanded, and phone companies were allowed into the TV business, the number of channels available to the average home began to rise sharply.
- In the late 1990s, DVRs became a reality, but growth was considerably slower than VCRs. It would take another decade before they were even in 20 percent of the country and needed to be addressed in developing television samples for audience measurement.
- Prior to 2005, changes to the media landscape – what was available to view and how it could be viewed – were gradual. Most people wound up getting the same access to almost everything. The slow pace of change made predicting consumer television viewing habits relatively simple, and slow-to-change research companies didn’t have to pivot too quickly, nor innovate too often.
- Since then, change has been rapid and constant – from the introduction of Apple’s first video iPod in 2005, to social media led by Facebook and Twitter in 2006, to the first iPhone in 2007, to Netflix and Hulu in 2008/2009, to the first iPad in 2010, Netflix’s first original scripted hit (House of Cards) in 2013, other subscription video on demand services such as CBS All Access in 2015 and the upcoming Disney, Apple, and Warner Media streaming services, among others. Today, the average home can receive more than 200 channels and at least one streaming service.
Some things worth thinking about:
- Television is a fundamentally different medium to DVR owners than to those without one, even when both groups are watching live TV – and there are likely dramatic differences in how much they are exposed to and recall commercials. But I have seen precious little research even attempting to understand these dynamics.
- How is the Adult 25-54 viewer who watches 10 hours a week of The Big Bang Theory on TBS or Criminal Minds on ION different from the Adult 25-54 viewer who watches 10 different programs over the same 10-hour period? Are viewers who can be reached over and over again on the same cable network any different from viewers who can only be effectively reached via multiple venues? Do they relate to advertising differently? Is advertising wear-out completely different?
- Is there a difference between Netflix, Hulu, and Amazon Prime viewers, or those who subscribe to all three?
- Over the past 15 years or so, a new generational divide has developed based on media device ownership and video platform viewing.
The youngest members of The TV Generation (those born between 1950 and 1969) are turning 50 this year. They, along with the generation that followed them, The Multi-Channel Generation (born 1970-1989), the first group to grow up with remote control and cable, do use multimedia devices and subscribe to streaming services, but they still see the TV set as their main (and often only) video viewing platform.
If you walk into my 20-year-old son’s room and observe his media usage (he’s part of the Multi-Platform Generation, born between 1990 and 2004), you’ll likely see him watching something from Youtube through Apple TV, a basketball game on his phone, and a soccer game on his laptop (if he’s not playing some videogame). He streams more than he watches on traditional TV – he is used to fast-forwarding though commercials and streaming commercial-free TV, and finds TV advertising to be more intrusive and annoying than previous generations. It is much more difficult, (if not impossible) for a Nielsen to fully measure his TV viewing than it is to measure mine. It will be even more difficult to measure total video viewing of The Mobile Generation (born 2005-). Yet even the most recent generations still see the big screen as most desirable and their first choice for viewing, but they are more likely to use other screens at the same time or if more convenient (e.g., my son will often put his phone on the table when we go out to eat so he can watch a basketball game – I never think to do that even though I have the same capability).
Change is happening quickly, and research needs to turn on a dime. Let’s not forget that no one cared much about researching what C3 actually measured back in 2007 because it was designed as a one- or two-season band-aid until pre- and post-buy computer systems could incorporate minute-by-minute ratings. That was 11 years ago, and we are still stuck with C3 and the equally meaningless C7. Few people, even at Nielsen, actually know what’s included in these so-called average commercial minute ratings. That’s what happens when one research company controls the marketplace currency.
Earlier this year, CBS publicly pretended to consider dropping Nielsen. I (and anyone who’s been in TV research for more than a few years) knew that would never happen, because the industry is too locked into the research company. It’s a monopoly, albeit and industry-mandated monopoly (at least on a national basis). The buying and selling of commercial time is so expensive that without an established entity like Nielsen keeping score, people fear the chaos that would ensue. The feeling has always been that even if Nielsen data is not “accurate,” it is probably directionally correct. And for the buying and selling of commercial time, directional accuracy and ratings that are stable enough to project out to the future counts just as much as precise accuracy. And as long as the ad-supported networks lock themselves into long-term Nielsen contracts, not much is likely to change.
Are Samples Still Able to Accurately Measure the Video Universe?
As viewing options expand and start to become substantially different among demographic cohorts, samples become less capable of measuring them.
In the olden days of television measurement (pre-2007), I believed single-source measurement was the way to go. Now, I do not think one sample or one company can or should be the gold standard for all-platform measurement. The things that go into developing a representative national sample are quite different for traditional television than for online viewing, and even more different for streaming.
The whole purpose of Nielsen’s national TV sample is that you and your cohorts, based on a dozen or so categories that Nielsen has determined to impact TV viewing (income, geography, presence of children, etc.) have similar viewing patterns.
This worked quite well from the 1960s (when three broadcast networks accounted for more than 90% of all TV viewing) through the 1990s (when the average household could still only receive about 30 channels and could only watch TV content on a television set). If I watched Seinfeld, for example, and was later in any gathering of my friends or colleagues and started talking about “being the master of my domain,” virtually everyone knew what I was talking about. My viewing was indeed remarkably similar to that of my “cohorts.” Today, in that same gathering of my friends and colleagues, quoting from some of my favorite shows – let’s say, The Good Fight (CBS All Access), Jessica Jones (Netflix), The Walking Dead (AMC), and Game of Thrones (HBO), would likely solicit blank stares from three-quarters or more of the people in the room.
These days, simply because of the number of viewing options available, differing device ownership, and varying levels of streaming access, examples of you and your demographic cohorts watching different things are much more commonplace. And as more and more things become available over the next few years, even fewer demographically similar people will watch the same stuff.
Nielsen samples are uncanny in their accuracy in measuring total television usage for broad demographic segments. In 2007/2008, I was involved in spearheading the Council for Research Excellence’s landmark Video Consumer Mapping Study as co-Chair of its Media Consumption and Engagement Committee. One of the study’s lesser-known findings was that Nielsen’s overall usage levels of households and broad demographic segments, such as Adults 18-49 were remarkably similar to the observed behavior in our sample. Once you started to look at narrower age groups, however, the reported Nielsen data started to stray significantly from the observed viewing data. And this was just overall TV usage levels, not individual program ratings (and was more than 10 years ago, before video viewing was nearly as splintered as it is today).
More recently (5 years ago), when I was head of Research at ION, we would often ask Nielsen how it was possible that an hour-long show like Criminal Minds can possibly lose half of its adult 25-54 audience in the second half hour and gain 40% among adults 18-24? The next day it might be the exact opposite. They would diligently analyze the data and then tell us that 10 people in the sample switched to something else, and that resulted in a reported rating decline of 48%. There was really nothing we could do about it other than make Nielsen do the same analysis every time we noticed some audience fluctuations that defied common sense. This is something certainly not unique to ION. If you scrutinize the daily demographic cable ratings of virtually any network, you’ll see many “illogical” fluctuations. Small ratings and small samples contribute to greater statistical error. If I were at a cable network now, I would favor almost anything that would significantly increase sample sizes.
Previous methods of drawing samples need to be revisited, as do how we evaluate media usage even among standard age groups. Consumers, whose media usage was once relatively similar if they were in the same demographic categories, are no longer as cohesive you might think. Large groups of broadly similar consumers now get and watch their video content in substantially different ways.
Neither can we continue to simply lump people with dramatically different media access and viewing habits into standard traditional demographic categories (such as 18-49 or 25-54). You will get answers, but they will often be less meaningful averages that miss what real folks are actually doing.
TV Ratings – To Be or Not to Be More Granular
I remember being at the 2007 meeting with Nielsen and the industry’s top national TV buyers, sellers and researchers, where we decided to officially move to C3 measurement. C3 was ostensibly designed to be a temporary band-aid until the agency pre- and post-buy systems were able to produce individual minute ratings (which they were able to do the following season). I remember pointing out at the time that the danger was giving the impression that we were getting closer to real commercial measurement, when we might actually be getting further away. Once the industry went through the chaos required to transition from average program to C3 ratings, we would not be inclined to do so again anytime soon (which, unfortunately, turned out to be the case). Anyone who remembers the craziness of that year’s upfront season, knows what I mean.
The problem of whether and how to move from C3 to more granular measurement is one of pitting the ideal against the practical. Advertisers and their agencies have long claimed to want nothing less than individual commercial ratings. But individual commercial ratings would be virtually impossible for agencies to deal with. It would require adding significantly to staff at a time when most research groups at agencies and networks have been sharply cut back. There are so many factors in the rating of a specific commercial (or even a specific minute) to make projecting future performance all but impossible – not the least of which is the lack of stability of such a small unit of time. While you can estimate how Grey’s Anatomy will perform on a program or even C3 basis, try predicting how it will do at 9:43pm for the third commercial in a seven-commercial pod.
A much better solution would be moving to commercial pod measurement for the buying and selling of commercial time. This starts to overcome the single commercial instability issue by averaging a string of consecutive minutes, and overcomes the C3 inaccuracy problem by accounting for the location of your commercial within the program. It also overcomes the estimating problem, since it would be relatively easy to estimate C3 as is currently done, and simply develop indexes for pod 1, 2, 3, etc. We could then still have individual commercial measurement to tell you how your specific commercial actually performed, but commercial pods would be used for buying and selling (and guarantees).
People don’t fast-forward through commercials because they dislike any individual ad. People fast-forward because they are able to, and don’t want anything interrupting their program viewing. So, if one ad is “better” than another, it will have virtually no impact on commercial skipping or commercial ratings.
Cross-Platform Measurement
While there is no question that we need to get cross-platform measurement right, it is still important to acknowledge that in-home television viewing on a television set is still by far the largest piece of the viewing pie. And we still don’t have that right.
While declining each year, roughly 80% of adult video viewing is still to traditional TV. Even among adults 18-34, traditional TV still accounts for 60% of all viewing. It’s about 75% among adults 35-49 and 90% among adults 50-64 (even higher among 65+). These figures wil n doubt be changing over the next few years as we see new major entries in the streaming wars. An upcoming issue of The Sternberg Report will look more closely at what needs to be done I the area of cross-platform measurement.
How to Measure the Accuracy of Current Research Techniques
When I was on The Council for Research Excellence, I proposed a methodology for gauging how accurate Nielsen was at measuring an increasingly splintered video viewing environment. Nielsen, of course, didn’t want to have anything to do with it.
At the time, the Council consisted of 40 of the top researchers from all sides of the industry (broadcasters, cable networks, local stations, media agencies, etc.). These folks are adept at dealing with an analyzing extremely detailed information. My suggestion was for Nielsen to meter each of our households. We would then detail our media exposure and viewing for one day (down to the second) and then compare our actual viewing to what Nielsen reports.
Given today’s media landscape, I would modify my proposal. Have Nielsen, ComScore, and a few other research companies meter the households of 50 or so of the top researchers in the industry. Set up numerous scenarios for what and how they will view video content during a single day. Then compare actual viewing to reported viewing. This will tell us what we need to do next. I’d be more than happy to be involved in the project.