There were a number of reasons for this, ranging from a lack of resources (in many campaigns, the media buying was actually handled largely by college-age volunteers, who weren’t about to try to reinvent the wheel); to the fact that political candidates themselves didn’t want to deviate from established practices (i.e., they liked seeing themselves on mass-reach broadcast television rather than on more targeted cable channels or online, regardless of whether these other media represented a more efficient way to target voters); to the fact that political media buying agencies get paid based on how much they spend, and thus don’t have much incentive to try to target more efficiently.

I was reminded of this regretfully abandoned project (perhaps I’ll resurrect it one day — still seems like an area that hasn’t received sufficient research attention) by this article about the current state of affairs in political media buying.  This piece illustrates how the process of buying potential voters is employing many of the sophisticated audience analysis tools and data that now characterize the process of buying potential consumers.  And, perhaps most important, this articule illustrates a growing emphasis on trying to reach potential voters more efficiently and effectively through online channels.

The final paragraph of this piece, however, reminds me of the findings that were emerging from my research five or so years back:

“Will 2012 mark a turning point when political campaigns realize that digital advertising offers them even more efficient and impactful marketing channels, and at much lower cost per impact, than their more traditional channels? Not in the sense that campaigns will invest less in their traditional go-to channels, as confirmed by the heavy TV spending we’ve already seen in Iowa, New Hampshire, South Carolina and Florida.”

The 30-second broadcast television spot still tends to be the dominant form of campaign communication.  The bottom line, I think, is that no candidate wants to cede that space to his/her competition. And so while efforts to target and reach potential voters online continue to develop, we still see this tremendous reluctance to diminish their emphasis on a communications strategy that seems a heck of a lot less efficient.  Plus, of course, it still tends to be the older Americans who vote in the greatest numbers, and these potential voters remain reachable via traditional broadcast television.

In the end, as much as new media technologies and analytical tools are dramaticaly changing the dynamics of reaching voters, these developments are taking place alongside — rather than replacing — traditional approaches.

The article describes some well-known examples, such as running a popular program a couple minutes over into the next program’s time slot (in order to goose that program’s reported rating), or labeling certain episodes of a program (specifically, those expected to perform below average, for whatever reason) as “specials” in order to keep them from counting toward the season average.  Obviously, these strategies are intended to exploit the specific procedural dynamics associated with how the ratings data are reported.

Some of the newer examples (at least to me) involved the distribution of the commercial loads within individual programs.  Specifically, networks will front load all of the program’s national commercials.  As the article notes, “Shows receive national ratings from Nielsen only up to the point when the last national commercial is broadcast — after that, the numbers simply do not count.”  The article doesn’t make this explicitly clear, but I’m assuming this strategy is a direct response to the adoption of the C3 rating (which functions as currency at the national level, but not at the local level).

This is a reminder that every new audience measurement system brings with it a variety of challenges — one of them being figuring out how best to exploit the particulars of how the system works.

What is particularly important about this, as this article  points out, is that “This marks the first deal MPG has done using Rentrak data as a guarantee — a watershed for Rentrak, which has said the TV market has room for more than one data set.”  In this case, then, the Rentrak data are standing alongside the Nielsen data as a currency.

This development reflects one of the key arguments in Audience Evolution, that the fragmentation of the media environment (not just in television, but virtually all media) is putting a tremendous strain on traditional currencies (particularly those derived from panels), and creating an environment that is much more receptive to multiple performance metrics/currencies operating simultaneously.

The key reason that we’re likely to see this pattern persist harkens back to Chris Anderson’s well-known Long Tail phenomenon. As Anderson notes, in today’s media environment, niche content providers are so numerous that, in the aggregate, their audience exceeds that of those content providers that attract traditional mass audiences. 

In the realm of audience measurement, this translates into a level of financial significance, in the aggregate, for the niche content providers that enables them to command the attention of audience measurement firms seeking to serve their needs — needs that are distinctly not met by traditional panel-based measurement systems, which are notoriously bad at measuring narrow appeal, niche content providers relative to those content providers that attract larger audiences. 

Correcting this imbalance is central to the appeal of set-top box data, as well as to a wide range of emerging audience measurement systems (discussed in Audience Evolution) that are seeking to measure dimensions of audience behavior beyond exposure.

In this context, the kind of bubble I’m referring to is similar to, yet different from, what we saw in the housing market, or the late 90s Internet, at least in terms of the possible underlying causes.  In the case of audience data, I’m wondering if it is possible that, a few years down the line we will find that any or all of the following have come to pass:

1) That the scope, depth, and constant flow of data so far eclipse the resources and abilities have organizations to make effective use of it that the market for such data ends up contracting dramatically (basically, the supply of data so far outstripping analytical resources and abilities — essentially information overload).

2) That the organizations that presumably could benefit the most from such data ultimately do not undertake the massive changes in organizational culture and personnel necessary to fully institutionalize the full and extensive use of such data into their operations (thus, that it essentially wasn’t possible for the influx of data to be utilized to optimal effect given the circumstances surrounding this data influx; I see the causes here as slightly different from those associated with item 1).

3) Or, perhaps,  a few years down the line, the growing array of data sources are being mined to their fullest potential and, it turns out, the benefits that accrue from such efforts don’t quite justify the substantial costs (in this scenario, folks were basically paying more for the data than the data turned out to be worth).

I’m just speculating here, of course. And this speculation was prompted by primarily by the pattern I’m seeing, where there’s lots of data being produced and purchased, but not always with a clear sense on the part of the purchaser (at least as far as I can tell) in terms of exactly how these data are going to be integrated into their business.  Similarly, some data providers seem to be aggregating certain types of data simply because they can — and don’t always have (or can’t clearly articulate) a clear idea of exactly how these data would be useful, and to whom.

I’m reminded of the late-90s Internet bubble, to some extent, where eveyone knew that they needed to be online, but weren’t really quite sure what to do once they got there. 

That Internet bubble would probably fit pretty well within the notion of technological determinism, a term used to reflect the belief that technology drives social, cultural, and institutional change; as opposed to technologies being somewhat subservient to the social, cultural, and institutional conditions in to which they are introduced.

I guess what I’m getting at, then, is do we need to worry about the presence of ”data determinism”?  That is, is it possible that the media sector, broadly defined, is assuming that data are capable of exerting greater power and influence over organizational performance than extent cultural, institutional, and economic circumstances will ultimately allow?

The Winterberry Group study, for instance, finds substantial “process and data structure challenges,” as well as challenges related to “rigid ‘silos’ and the paucity of data-savvy marketing operations, IT and sales talent.” 

Is it a foregone conclusion that issues such as these get resolved? And if they don’t, do we see a bit of a contraction from what currently seems to be a bit of a gold rush mentality surrounding the supply and demand dynamics of audience data?

Consider, for instance, that last month, set top box data from audience measurement firm Rentrak were used by Viacom in an effort to resolve an ongoing dispute with Nielsen over a sudden drop in ratings for its Nickelodeon children’s network.  The Media Rating Council, which is investigating the dispute, is comparing Nickelodeon’s Nielsen ratings with Rentrak set top box data, as well as with TiVo’s set top box data. Apparently, only the Nielsen data show a dramatic drop in Nickelodeon’s ratings. 

The fact that the MRC apparently considers set top box data to be a meaningful basis for comparison in assessing the accuracy of Nielsen’s panel data represents an important stepping stone in the validation of STB measurement (somewhat ironic, considering that the MRC has yet to formally accredit any STB measurement services).

At the same time, STB services seem to be making inroads amongst both buyers and sellers in the audience marketplace.  A couple of weeks back, the Post-Newsweek television station group signed a deal with Rentrak.  Of particular significance is the fact that the station group plans on using Rentrak’s set top box data as currency for deal making.  Also quite recently, large media buyers such as Camelot Strategic Marketing & Media, and Carat have signed deals with Rentrak.  As I mentioned in an earlier post, Rentrak’s recent offering of “exact” commercial ratings (that is, the actual rating for the specific commercial) holds substantial appeal for media buyers dissatisfied with Nielsen’s average commercial ratings (which is the average rating of all of the commercials that air within a particular program).

Not everyone, though, is optimistic about STBs being able to replace people meters.  A couple of recent pieces by MediaPost’s Gary Holmes (one dealing with the use of STB data at the national level; and one dealing with the use of STB data at the local level) shed a very skeptical light on the prospects for STB data, via a very comprehensive assessment of the complex set of technological and institutional factors that impact the introduction and diffusion of new audience measurement systems.

Holmes predicts that the likely outcome is some form of a hybrid panel/STB system (something Nielsen is currently working on). And, given the way that online audience measurement is quickly moving toward a hybrid approach (utilizing both panel and site-centric systems), it seems reasonable that a similar path will be followed in television.  The bottom line is that today’s electronic media environment (regardless of platform) represents a degree of fragmentation and complexity that exceeds the capabilities of any one measurement approach.

I planned on summarizing the basics of my presentation here, but fortunately one of the faculty members in attendance, Sasha Costanza-Chock liveblogged the presentation, as well as the Q&A.

Further, you can find the PowerPoint slides that accompanied the presentation here.

The goal of this initiative is to bring some standardization to a measurement arena that continues, even in 2011, to be characterized as “the Wild West.”  As one ad exec quoted in this story about the 3MS initiative’s recent activities points out, ““Look at four different measurement services and you’ll get four radically different pieces of information.” 

This kind of inconsistency across measurement systems breeds two things that are anathema to participants in the audience marketplace: complexity and uncertainty.  This quote from a Universal McCann executive is particularly telling: “To spend $100,000 in online advertising is far more complicated than $10 million in television.”  This statement is a reminder that the shift in advertising dollars from traditional media to online platforms (which is progressing steadily) could be happening much more quickly if the systems of measuring online audiences were further along in their evolution.

The 3MS group has, to this point, announced five guding principles for digital measurement.  These are: 1) to move toward a standard of viewable impressions rather than served impressions; 2) to move toward a currency based on audience impressions rather than gross ad impressions; 3) the creation of a transparent classification system; 4)  identifying and defining metrics most valuable to brand marketers and defining and implementing reliable standards for the existing metrics; and 5) making digital measurement increasingly comparable and integrated with other media.

As this list makes clear, online audience measurement still has a long way to go before it’s satisfying the needs of advertisers. And, it’s worth noting that the pressure is on online audience metrics to look more like those used in traditional media.  This is a potent reminder of the extent to which the existing structures and practices in any media system exert a powerful influence over the evolutionary path followed by new entrants into these systems.

A couple of recent developments highlight the gradual, evolutionary process that characterizes even a change that seems as sudden and dramatic as the adoption of the C3 currency in 2007.

First, it is important to emphasize that the C3 rating has applied, up to this point, only to the purchasing of national television audiences.  As I’ve discussed previously, the C3 has not yet trickled down, so to speak, to local television markets; and it is only beginning to tentatively appear in the measurement of online audiences.  Nor has it yet been applied to other viewing platforms such as on-demand. 

The C3′s reach may soon be expanding, however, with the recent announcement by Nielsen that it is developing an “On Demand C3.”  The ODC3 system will produce a single rating that includes a show’s live telecast, along with time-shifted DVR viewing (the components of the traditional C3), and a combined C3 for all of the older episodes available on demand.  Of course, as is the case with the TV/online combined C3 metric I discussed not too long ago, it is required that each program stream contain exactly the same commercial load. 

In other C3 news, the very nature of the C3 may be continuing to evolve.  The current C3 rating is calculated by computing the average rating of all of the commercials contained within a particular program.  Some stakeholders have criticized Nielsen for this lack of granularity in the C3 (why, they argue, shouldn’t they be able to find out the audience size of an individual commercial, or at least an individual commercial pod?).

This may be changing with the recent announcement by emerging Nielsen competitor Rentrak that the company will be offering “Exact Commercial Ratings.”  These ratings, which will be derived from Rentrak’s digital set-top box measurement system, will allow subscribers to assess the ratings performance of individual commercials.  However, in keeping with the persistent pattern we see in audience measurement, this more granular approach to understanding television audiences will only be available at the national level and not at the local level.

One could argue that these efforts by a relatively small competitor to Nielsen hardly herald a change in the nature of the dominant currency in the television industry.  However, history tells us that, while Nielsen might not always take the lead in innovation in audience measurement, they do tend to respond to (and absorb) the innovations produced by others (for example, Nielsen didn’t introduce the People Meter in the U.S. until Audits of Great Britain brought it over from the U.K. in an effort to crack the U.S. market).  Thus, it stands to reason that Nielsen will soon respond to this development by Rentrak with its own form of exact commercial ratings.

Traditional ratings analysis approaches to audience research have yielded a wealth of valuable insights into the patterns of audiences’ media usage that have helped in the formulation of strategies and tactics for for attracting, retaining, predicting, and valuing audiences.  This research has taken into account a variety of factors in order to better understand audience behavior, ranging from content characteristics (program type, quality, etc.) to structural factors (scheduling, competition, range of choices, etc.) to audience characteristics (e.g., demos). (For vital overviews and assessments of this research, see two books by Jim Webster and his colleagues, The Mass Audience and Ratings Analysis.)

So, needless to say, those of us who do audience research need to start thinking about the structural, audience, and content factors that can help to explain and predict how individual progams will perform on contemporary criteria such as engagement, involvement, buzz, etc.  I’m starting to try to formulate some thoughts along these lines and to (most important) obtain some of the relevant data from some of these measurement firms in order to initiate this line of research.

Below are some questions I’ve formulated thus far in trying to outline some initial directions for this research agenda.

Content Factors:

What is the relationship between various program types and engagement, involvement, etc?

There seems to be reason to believe at this point that reality programs tend to perform better than scripted programs; but are there more extensive systematic program type effects? Sitcoms v. dramas? Are there measuers of quality that might be useful for understanding audience engagement/involvement? I’m thinking, for instance, of the program testing that networks have long conducted that has proven to be a somewhat helpful predictor of a program’s traditional ratings performance?  How do these pre-tests perform in terms of predicting the engagement and involvement indicators being provided by measurement firms?

Structural Factors

What is the relationship between scheduling factors and engagement, involvement, etc?

Certainly, in a DVR/On Demand era scheduling is less of a factor, but I bet scheduling still plays a role in the level of engagmenet/involvement demonstrated by audiences.  To what extent is engagement/involvement a function of dayparts?  Do we see lead-in or lead-out effects?  That is, might one program’s engagement level be a function of the engagment level of the program that precedes or follows it? Might engagement/involvement levels be a function of the day of the week? (maybe we’re more or less engaged with programs we watch during the week vs. the weekend).

What is the relationship between consumption platform and engagement, involvement, etc?

This is obviously a new level of analysis that really didn’t concern traditional audience researchers.  And I don’t know yet the extent to which the emerging audience measurement services are able to parse out their data according to consumption platform.  But one can certainly imagine that the nature of the platform used could impact the extent to which an audience member feels/demonstrates engagement, involvement, etc.  Might engagement/involvement, for instance, be a function to some extent of whether the content was consumed via a live versus an on-demand platform?

Audience Factors

We already have a fairly good idea that, in terms of the measures of audience engagement and involvement that are being introduced into the marketplace, program performance is very much a function of the demographics of the program’s audience.  Older television viewers, for instance, are just not as likely to blog and tweet about their favorite programs.  Recent research by Nielsen provides some demographic data on the types of people who are more likely to discuss television programs online. No real surprise — this activity tends to skew young and male. 

However, learning more about these patterns and their relationship to individual programs — or program types or networks — could ultimately provide very valuable insights into how the representation (and under-representation) dynamics of these new measurement systems might mirror or deviate from the representation and under-representation dynamics present in traditional ratings panels.  Such patterns obviously can have dramatic implications for the nature of the programming that is produced.

Pandora CEO Tim Westergreen asserts that it is the lack of a “common currency” in the audio media space that is contributing to the company attracting a share of the pie of audio advertising dollars that is well below the company’s estimated 4% share of the overall music audience.

What is particularly interesting is Westergreen’s emphasis on the need for a common currency that allows for cross-platform comparisons and aggregation of audio/music listenership across the full range of relevant platforms, including computers, mobile devices, satellite radio, and terrestrial radio.  As the TV Everywhere discussion highlighted, though, there are often a variety of reasons why certain stakeholders would resist a common currency, particularly if the CPMs that can be obtained on some platforms are significantly higher than the CPMs that can be obtained on other platforms.

And, as we typically see in the realm of audience measurement, the demand for the expansion or improvement of measurement systems and/or new currencies needs to be fairly widespread across the relevant stakeholders (particularly the largest stakeholders, who pay most of the freight when it comes to audience measurement services).  This is in fact one of the key ways in which incumbents are able to slow the development of new media technologies, services, and platforms — by resisting the development (and by resist I mean by refusing to cover the costs) of measurement systems that could provide a more level playing field between old and new media platforms.  As I discuss in Audience Evolution, we’ve seen this pattern in radio; we’ve seen it in television; and we’re starting to see it online as well.