24 March 02005

What does On-Demand Media really mean?

This week Arbitron and Edison Media Research published a report of their research survey on Internet and multimedia usage, The On-demand Media Consumer. The headline result being quoted is that "One in ten Americans show a heavy preference to control their media and entertainment". Reading the summary report, however, suggests that this conclusion is not warranted by the data.

This article covers the flaws in interpretation in the report, and then suggests some distinctions that might form the basis for a more sophisticated approach.

The biggest problem is that the researchers constructed a scale which they say "represents the level of control that consumers exercise on their own media usage", and most of the report is based on correlations with this scale. If you look at the fourteen items that were used to make up this scale, it's clear that these are just measures of adoption of the latest gadgets and technologies (e.g. owning a BlackBerry®, or a portable DVD player, having watched or listened streamed video or radio online in the past month, spending seven hours or more on the Internet per week). Not for these researchers the methodological rigours of testing the reliability of their scale using Cronbach's alpha or other established disciplines.

The technologies included in the scale provide users with different levels of control. iPods and TiVos may provide almost total control, but listening to Internet radio does not necessarily provide direct control (especially if listening live), just a wider selection of programmes. It is true that the latest media technologies do in many cases give their users more control over what is played when, but that is not the only way the new developments distinguish themselves from older offerings. So engagement with the latest gadgets in services is not a straightforward index of 'level of control'; it just shows whether you're an early adopter or not.

Why have the researchers confused these issues? It looks like they are targeting their report and their services at the advertising sector that provides revenue for many areas of the media. And I guess they are trying to grab the attention of their target audience by suggesting that increasing use of on-demand technologies requires advertisers to re-think their approach to placing their messages in different media.

Once you establish that the scale in the report represents adoption of technology, and not control over media usage, several of the other results become unsurprising. For example, the report says that people who score highly on their scale are 'early adopters'. Well, since two of the fourteen items on the scale are "first to try new products or services" and "try new products or services before most", that comes as no big shock! High scorers are also less likely to be over 55, are more educated and "upscale" than other Internet users, and more likely to have multiple computers at home. Again, these are well-established demographics for early adopters.

The report has an interesting table showing which devices and services have high percentages of owners who say they "love" them. The report interprets the data as showing that on-demand devices have high enthusiasm among their users. The fact is that it's a staple of usability and user satisfaction that users like to feel themselves to be in control of their devices — see, for example, rigourously developed scales of multimedia usability like MUMMS. This is why products renowned for their simple, user friendly design (like the iPod) score higher in the table than other products in the same class, even though the latter ostensibly offer the same level of control and 'on-demand-ness' as the former. And it's also probably the reason why old-fashioned over-the-air radio scores twice as high on the "love" ratings as new, on-demand Internet radio. The problems that Internet radio occasionally has with stalled networks and so on makes users feel less in control of it than AM/FM radio, which works reliably, even if it doesn't offer time-shifting flexibilities.

There is one other finding about terrestrial radio in the report that sticks out like a store thumb amid the interpretation given to the rest of the data. Eighty two per cent of Americans say they will "listen to terrestrial radio in the future as much as they do now despite advancements in technology". Which is surely a ringing endorsement for not-on-demand media however you look at it.

To understand and study the spread of on-demand media we need a more detailed analysis that distinguishes differing degrees of control over media 'streams' and how coarse- or fine-grained the control is. For example, thinking of music audio streams, there is a spectrum from:

  • terrestrial radio — no control over programming, track selection or scheduling;
  • digital radio with the Bug — limited 'micro' control of scheduling through ability to pause, rewind and record live radio;
  • Internet radio with time shift capability (e.g. BBC Radio Player) — control over what programmes you listen to and when, but no control over programme contents;
  • Internet radio with interactive programming (e.g. Last.FM — track selection can be rejected by users, and users can also determine how closely the programming is tailored to their profile of preferences (from 'personal radio' to 'random');
  • MP3 player in shuffle mode — user has total control over the library of tracks, but no control over sequencing; to
  • MP3 player in playlist mode — user has total control over selection and sequencing.

A similar spectrum could be developed for video media streams, but would have to distinguish between the more varied and complex set of 'units' in video — i.e. complete film, TV programme, scene selection, or music video track, as opposed to the basic 'track' unit that applies in popular music.

If you start to research media use at this more detailed level, I believe it will be possible to understand consumer preferences in a way that resolves any apparent inconsistency between a desire for greater user control on the one hand, and a continuing commitment to traditional media formats like radio on the other. Previously I've argued against the claim (also based on questionable research) that iPods will spell the end of radio on the ground that consumers are likely to want a mix of control over 'known' media and an introduction to 'novel, unknown' media that might enrich their collections.

Playlisting satisfies the first desire, and is a true on-demand, fully user-controlled mode of consuming media. Satisfying the second desire is more complex. People like the convenience and simplicity of traditional radio: in a way these features provide users with a degree of simple-flick-of-the-switch control. But new technologies provide other ways to give control over the discovery process. These include:

  • the massive increase in the programming available — through digital, satellite and Internet distribution — which provides greater possibility for consumers to find channels that match their niche interests;
  • timeshifting, which again gives consumers more flexibility to listen or view what they want when they want, rather than being at the mercy of broadcast schedules;
  • interactive recommendations, which — as in the Last.FM example given above, or Amazon's filter-based recommendations — learns consumers' preferences and presents media that appear to match those preferences.

Research that distinguishes between different levels (or 'grain sizes') and different types of control in media consumption will tell us more than research based on crude assumptions that all new devices and services provide more control than traditional media outlets.

Posted by David Jennings in section(s) Human-Computer Interaction, Ideas and Essays on 24 March 02005 | TrackBack
Post a comment

Remember personal info?