The New York Timesb Stuart Elliot covered NBCbs marketing tactics (a lot of radio).B San Francisco Chroniclebs Tim Goodman opined on production studiosb response to Leno, how theybll wish disaster on NBC as punishment for a strategy that marginalizes their contributions to the television ecosystem.B Fellow Jack Myersb MediaBizBlogger and media curmudgeon Charlie Warner has a three-part series on what hebs learned from the premiere.B Other outlets have relayed The Leno Showbs ratings, commented on the above commentary, and foretold further disruption as the competing networksb prime time programming comes online.

The days after a show has premiered in its new timeslot are an opportunity to observe how television viewers have adjusted their choices of what to watch.B Focusing on a particular new program, for those viewers who have chosen to tune-in, what kind of programming were they watching in the weeks leading to the premiere and how many are demonstrating loyalty and tuning in multiple times?

Looking at those viewers who tuned in to the first week of The Jay Leno Show reveals indicators of some program preferences that youbd expect and some that would surprise.B The two charts below examine the programming airing at weeknights at 10PM that the Leno audience viewed prior to the September 14 premiere.

The first chart reveals televisionbs emerging long tail.B The chart shows the percentage of the Leno audience that watched any of 650 different programs airing at 10PM on weeknights across broadcast and major cable networks in the weeks of August 31 and September 7, ranked in descending order by the percentage of the Leno that tuned-in.

Only 10 of the 650 programs attracted 5% or more of the audience that went on to view at least one episode of The Jay Leno Show during the week of September 14.B Of the 10, 9 of the programs were on NBC.

The next 30 programs each attracted 1% of more of the Leno audience.B Half of those next 30 programs aired on ABC, led by 20/20 and Primetime news programs.

The second chart examines the relative likelihood of the The Jay Leno Show audience to have tuned in to various programming in the two weeks leading to the premiere.B It shows the top 25 programs by Rating Index, a ratio of the program rating among viewers of at least one episode of The Jay Leno Show and the overall program rating.B A Rating Index value of 100 indicates that the Leno audience was no more or less likely to have viewed the program as the general viewing population.

The top 8 programs by Rating Index are, as one might expect, NBC programs.B The audience that had tended to watch NBC at 10PM continued to watch after Leno premiered.

Interesting entries in the top 25 programs by Rating Index are Bravobs Flipping Out and PBSb Great Lodges of the National Parks and Wild River: Colorado.B Leno viewers were nearly 3 times more likely than the typical viewer to have watched those programs. The Bravo entrant to this list is likely the result of cross network promotion.B The crossover of audience genre affinity to explain the connection with Leno and the PBS programming is worthy of more scrutiny.

The degree of loyalty to The Jay Leno Show is similar to observations of loyalty to other programming: low.B The chart below examines viewer loyalty to Leno at 10PM.B Of the viewers who watched any of the Leno Show during the week of September 14, a majority, 65%, tuned in to one episode.B Only 5% of Leno viewers tuned in four times in the programbs premiere week.

Audiences arrive to a program, watch some part of it, and depart likely never to return.B Networks bownb a viewing audience the same way that you bownb the breeze that comes through your homebs open window.

Viewers sampled programs before The Jay Leno Show aired at weeknights 10PM on NBC, and viewers will sample programs after the September 14 premiere.B To make the most of the rest of the Fall 2009 season, program marketers will do well toB understand which viewers have a proclivity to sample and which viewers have a proclivity to return to programs.

Consequently, the types of programs that broadcast networks can air, and the subsequent growth rates of these programs, differ considerably from the programs and program growth rates on cable.B Broadcast networks tend to only air shows that are likely to earn high ratings in the first season, leading to programming that sometimes has a one-size-fits-all feel.B Cable networks, on the other hand, can produce niche shows that might initially have low ratings but that have a chance to garner higher ratings several seasons down the road.

Cable programs, then, are much more likely than broadcast programs to exhibit significant season over season growth.B Most broadcast programs start out with large audiences that begin to dwindle each year, while cable programs tend to start smaller and progressively grow larger each season.

Take HBObs True Blood, USAbs Burn Notice, AMCbs Mad Men, and TNTbs Leverage for example.B All of these shows enjoyed a healthy growth with each additional season, with True Blood taking home the prize for the largest seasonal jump. B ABCbs Desperate Housewives, NBCbs Heroes, CBSbs CSI: Crime Scene Investigation and FOXbs House didnbt fare as well in that they lost audience with each progressive year.

Note: Data from Nielsen Media Research as reported by TVByTheNumbers & Wikipedia

B That is not to say that there arenbt a few anomalies on broadcast networks.B Both B CBSb NCIS and NBCbs The Office have grown by 50% (11.8M to 17.8M) and 70% (5.4M to 9.2M) respectively since their first seasons. B While those are impressive changes, the overall downward trend still remains for most broadcast networks.B In fact, of the shows we researched for ABC and FOX, we couldnbt find one that had grown in audience recently when comparing one season to another.B

As the options from cable networks continue to increase, it is harder for broadcast networks to maintain their grip on audience.B While broadcast networks still win in terms of audience size, cable networks are slowly catching up, producing numerous programs with large ratings, even if the ratings sometimes take a few seasons to develop. B With the Fall 2009 season quickly approaching, broadcast networks are hoping as usual that they have huge hits on their hands.B Cable networks also hope that their programs will be successful, but know that their shows likely have a bit more breathing room than broadcast programs, which tend to be dropped immediately if ratings are lackluster.

For the past three months in my new start-up, Simulmedia, we have had a team of data analysts, behavioral biologists, and statisticians using TNS Media Research’s InfoSys TV system to analyze anonymous, aggregated set-top-box data records representing recent viewing histories of approximately 350,000 households in the Los Angeles market to better understand how people watch television. The insights we’ve been able to glean already have been pretty extraordinary, but one really stands out: Television viewers watch significantly fewer episodes of each program that we had anticipated. Program loyalty is the exception, not the rule.

We selected a number of top shows from broadcast and cable networks and calculated a “loyalty score” for each program. Surprisingly, we discovered that a relatively small number of folks who watched a particular show this season, watched two or more episodes. The show with the highest loyalty score — “American Idol,” unsurprisingly –B garnered a loyalty of only 65%. Thus, only 65% of the folks in our sample who watched “American Idol” this season watched it more than once. “Lost” came in at 52%, “Gossip Girl” at 50%, and “Mad Men” at 33%. On average, only 46% of folks who watched a particular program in our sample watched two or more episodes.

Maybe this isn’t news to many of you, but we were surprised. Why does this seem to contradict conventional wisdom, largely based on panel- and sample-based research? Probably because panels can project, for example, that a particular episode had an audience of 10 million viewers, with 75% of those viewers from a particular demographic group — but they can’t tell you that two-thirds of those folks are different from those who watched the show the week before.

Of course, the elephant in the room is the DVR (digital video recorder). The set-top-box data we analyzed did not include DVR viewing. However, since neither Nielsen nor advertisers value that viewing very much — given DVR ad-skipping — the linear viewing reality that program loyalty is the exception, not the norm is also part of the economic reality of television today.

What does this mean for television companies and advertisers? I think it’s now incumbent on players in this market to dig into viewer data. Anonymous set-top-box data and analysis is now available from a number of sources and firms. There is no longer an excuse not to know what TV viewers really do. Now is the time to find out, and adjust business practices according to reality, not just history. What do you think?

(This post originally ran on MediaPost’s Online Spin on May 28th, 2009.)

Segmentation is a useful technique in our work to understand people’s viewing choices.B We divide and subdivide the general viewing audience into segments based on data describing people’s attention to television programming.B Then, we analyze how our segments tune in to different networks’ offerings and their responsiveness to different promotions.B B B

The analysis sensitizes us to trends in tune-in and promotional responsiveness.B It inspires ideas on how we can improve our segmentation and the classification of attention data that underlies the segments.B

Ultimately, we hope segmentation can help explain changes in tune-in.B Given how different segments tune in to a certain program, we want to make recommendations on the best promotional messaging strategy.B Or, when a program grows more popular, we want to look at how different segments tuned in and answer the questions of how and why did ratings improve.

In the meantime, through our sandbox, we want to share our early exploration into people’s viewing choices and get feedback on our approaches.B

As a foundation for taking in our exploratory work, we present our initial Genre Segmentation.B Genre Segments are unions of audiences who tuned in to similarly classified programs.B We’ve assigned a name to each Genre Segment corresponding, sometimes humorously, to the type of programming they’re likely to enjoy and written a brief persona description to help cement the relationship.B

The size of each Genre Segment population is a function of the definition we’ve issued to determine whether a viewer belongs to that segment or not.B In the table below, the size of the segment is indicated by the percentage of the viewing universe that belongs to the segment.B Segments with more restrictive rules are smaller; less restrictive, larger.B

You’ll find Genre Segments in our analyses of Audience Attentiveness and our Network Audience Heat maps.

B

Sure, we have a lot of data, and we are good at slicing, dicing, and analyzing it many different ways, but data alone does not a company make.B We believe that, in order to know your audience, you must also be your audience. B If none of us watched the types of television shows or movies that we would help promote, we couldn’t relate to the audiences we are looking to understand.B We would miss the connections that people have with television shows and movies.B B

To illustrate my point, I’ll share one of my connections.B

When I was six years old, my grandfather passed away.B I remember being at the cemetery hiding underneath my mother’s jacket while bagpipes played “Amazing Grace”.B I was crying my eyes out.B

As time passed, I moved on and didn’t really think about that day too much.B That was, until I saw “Tommy Boy.”B

Yep.B “Tommy Boy” – the 1995 comedy with Chris Farley and David Spade.B

Why “Tommy Boy”?B In the movie, Tommy’s father passes away and after the funeral, Tommy walks through the cemetery while “Amazing Grace” plays on the bagpipes.B When I first saw the scene, I realized I was crying but not because Tommy lost his father.B It was sad of course, but I had seen funerals in movies and television shows many times and not a drop.B I have also heard “Amazing Grace” on the bagpipes and nothing there either.B It was the combination of this scene with the song that, to this day, causes me to cry.B Not that there’s anything wrong with that.

So now what? What do we do with this information?B Is there something there?B Do people have connections to movies and shows that might help explain what they’ll watch in the future?B The answer is “I don’t know”, but, because we recognize the possibility that people have strong affinities for programs based on surprising and counterintuitive combinations of factors, we are thinking about this type of thing.

Please feel free to share your comments or your connections that you might have.