TV on Laptops or PCs: The Knowledge Gap
In the UK our staff are always being asked if we can measure the viewing of content on PC’s and laptops.
The answer is No, or rather Not Yet, at least not in the way conventional TV is measured. (There are some other options which I will mention later).
Traditional audience measurement needs credible, common standards that provide the basis for advertising trading currencies.
Audience measurement companies are very aware of the challenges of digital viewing. In the UK BARB is field-testing a device that will capture TV viewing on laptops.
The chart in this blog tells you about the take-up of viewing to TV content on computers. The test panel of 75 people set up by BARB uses a new meter from TNS/Kantar called the “Virtual Meter”. Albeit based on a very small panel, we can see that the use of laptops and PC’s for watching TV is growing fast.
BARB is also trialling a system for monitoring on-demand viewing. This requires much larger data sets than for conventional “linear” TV. For the system needs to recognize a digital “signature” on every piece of content monitored.
You may ask: if they have the technology, why don’t they use it? But the technology is in fact the least of the problems.
Measurement systems like BARB are based on sampling, on panels of people who represent a population. But statistics based on sampling need adequate samples, otherwise the errors can be very large and the findings very unreliable. (My colleague, Farid el-Husseini, has given examples of small channels in the BARB system where the errors on some audiences can be plus or minus 200%).
Those who actually watch content via PC’s or laptops do so, apparently, for nearly two hours a week on average. But across the population, that is, taking account of all the people who do not watch on the Internet, their viewing amounts to only a small proportion of total viewing, between 10 and 20 minutes per week — whereas the average person watches conventional TV for three or four hours every day.
Panels are expensive to run and expensive to recruit – and if they are not large enough to give a reasonable degree of accuracy for every item viewed they are not much use. The BARB panel covers nearly 6000 UK homes. In a perfect world, BARB would wait till enough of those panel homes were watching TV on laptops or PC’s to give a reliable rating.
Unfortunately, it is still too early to say whether or when Internet viewing to TV content will be reliably covered by systems like BARB or Nielsen in the US.
As for the other options, lots of companies carry heaps of information about the people using their own services because they own the servers on which subscriber viewing is captured. But they have been quite reluctant to publish detailed data or organize it in a way that makes it comparable to BARB statistics.
BSkyB in Britain, for example, uses its server data on household behaviour to create statistics comparable with BARB data but they are not in the public domain. (This kind of data is called “server side”, whereas data gathered from people with devices in their homes is called “client side”.)
Google Analytics, which many companies use to monitor hits on their websites, is a “client side” system because it relies on peoples’ willingness to carry a code that monitors where the hits come from. I expect some the traditional audience measurement companies are concerned that someone like Google may jump in ahead of them.
All this is frustrating for rights owners and producers and advertisers. Though they may be getting accurate information about the hits or downloads of their own content – because that is how they earn their living – they are not learning nearly as much about these new viewers as they would like and they have no way of monitoring competitors.
Naturally, we are very aware of this knowledge gap since our clients keep reminding us it matters.
We may have more to say on this before too long. Please watch this slot.