Amid the proliferation of channels, connected TVs and fragmentation of viewers onto new platforms, how can audience figures be accurately measured to stay relevant.
Whether anyone is actually watching is a question that has haunted broadcasters since the invention of television. The trouble is, it's getting harder to be sure just how many people really are watching. The proliferation of channels in many markets has diluted the already limited statistical credibility of ratings based on samples of tens of thousands of households within populations of millions. In the US, for example, it has been suggested that available ratings data may not be providing accurate viewing figures for at least 300 of the 400 national cable and broadcast TV channels. Second, viewing habits are no longer tied to TV schedules, with more use of DVR, catch-up TV, VoD and mobile TV services, not to mention the confusing, ever-expanding world of online video.
The industry is trying to solve the problem, with the likes of BARB in the UK and the Coalition for Innovative Media Measurement (CIMM) initiative in the US working on new measurement techniques. At the same time, audience measurement specialists and technology companies with backgrounds in related fields, including (in no particular order) Rentrak, TiVO, TNS, TRA, Canoe Ventures, NDS and Openet each continue to develop solutions allowing operators to measure consumer activity across various content delivery platforms.
A unified measurement system
But a unified measurement system that would enable broadcasters, advertisers and other interested parties to get a clear picture of who is really watching what and when, still seems a long way off.
“The greatest challenge in the short term is the increasing divergence of platforms,” says Steve Plunkett, director of innovation at Red Bee Media. “You've got TV manufacturers becoming platform providers and they all have slightly different systems. Over the next two to three years the rate of growth in online viewing is likely to increase because of connected TVs. What's lacking at the moment is any sort of uniform mechanism or standardisation. Individual platform providers or broadcasters are having to fix that for themselves.”
Nor will a unified system come cheap. “Everyone knows what they need to do, but it's just so expensive,” says Christian Brent, vice-president of strategy and audiences at BBC Worldwide. “The old ratings panels just aren't big enough to measure the fragmented landscape. It's now so huge they would have to be magnitudes of ten bigger and it's just not cost-effective to do that.
“A lot of the ratings panels are getting bigger, but smaller channels now often can't afford to buy ratings data. They have to go back to surveys and focus groups. It's almost like we're sliding backwards in terms of what we know about audiences,” says Brent.
There's also a question, he adds, over whether some of the mainstream channels, perhaps particularly in the free-to-air world, really share much enthusiasm for a unified system. “If the smaller channels aren't really being measured their share goes back to the bigger channels,” he notes. “Sometimes maybe the big terrestrial channels are dragging their feet because it's not necessarily in their interests to have a full picture of their smaller competitors' ratings.”
The problem with STB data
Return Path Data (RPD) from digital set-top boxes (STBs) will form part of the answer to the technical problem and is already used for this purpose in a number of markets. But it's not perfect. “The digital STB knows what you're watching, but there's the issue of the box not knowing who's watching,” Brent explains. “With the old ratings panel you would have a little clicker to say which person is in the room. Because of that we're used to having demographic ratings statistics. The RPD doesn't have that.
“The way that TNS [and other ratings data providers] are approaching that is to build some clever algorithms so they know who owns the STB, they know who lives there and they can predict who might be watching. But you're introducing some uncertainty.
“Then there's the problem of people switching off the TV at night but not the STB. It still thinks that channel's being watched. That's another place where you need algorithms to support the RPD. And of course they don't use data from all the STBs: it's just a sample. RPD is an interesting area, but presents a different set of problems.”
But RPD measurement is becoming more sophisticated, with the likes of NDS developing more advanced RPD software able to recognise more complex behaviours like consumption of time-shift TV. “We support the concept that there will be a gold standard, separate and independent metering system which establishes base parameters for mainstream TV viewing and that RPD is brought from a larger sample of boxes and devices and used to complement, extend and deepen the measurement,” says David Whittaker, director of business development and advertising technology at NDS.
“We've pursued this policy in the UK with cooperation from BARB and Sky, TNS is the data processing partner, AGB Nielsen the metering partner. It extends BARB reporting to include time shifting measurement, so the industry can see how much time-shifting and ad-skipping takes place. At the same time the technology is also used to develop more detailed understanding of the smaller and newer channels.
NDS has pursued this approach, of putting appropriate software into the boxes, in other territories around the world, including, in the US, DirecTV's view panel. “Most of the other arrangements are confidential, but we're currently working with Joint Industry Committees (JIC) in various markets,” adds Whittaker.
Concurrent is also consolidating measurement data from multiple platforms, developing answers to fundamental questions about how such data should be stored and managed. It claims its Media Data and Advertising Solutions (MDAS) now represent the first complete, intelligent solution for collecting and aggregating data from television, online and mobile viewers.
“It's all about the infrastructure, how you collect the data from STBs, VoD servers, interactivity servers; how you get that into a data warehouse, how it's stored, what's 'historic' data - all those questions,” says Paul Haddad, senior vice-president of media data and advertising solutions at Concurrent.
“How do you collect the data, how do you audit it, what happens if six of the 130 Comcast markets don't report data for last night because their servers went down? Do you report anyway, or wait until that data comes in? We think our approach will drive standardisation on how data gets collected and the interfaces. The same technology can do that with servers for the web; and from 2011 will do mobile too. But we don't yet pull all of them together.”
Clients include Charter Communications in the US, using the technology to measure VoD, interactivity, linear and DVR activity across three million STBs. “They need to report what's going on for advertisers, but we're finding that the more Charter are getting data from multiple services the more they're able to better manage their business,” says Haddad. “It's enabling them to better understand their customers.”
Wireless and cable business intelligence specialist Openet also offers software capable of collecting real time click streams from sources including linear broadcast and switched digital video, VoD, DVR and interactivity. “We've gone to the sources of the data: the servers and the STBs,” says Kelly Anderson, director of multi-channel video solutions at Openet.
“There's some legacy issues around the STBs, but operators are upgrading those or installing middleware. So that situation's getting better, but even without STB data we're getting it from other sources, such as switch digital feeds and overlaying that data to produce reports operators can use not only to see what the value of their content is, but also to measure those channels that Nielsen doesn't measure. So it allows long-tail selling for ad groups, they can utilise that data for content negotiations and look at what their packages need to offer users.”
But while these innovations are impressive, they still leave much online video viewing untouched. “When you get to the over-the-top (OTT) world the wheels completely come off,” says Brent. “There are experiments going on around the world, but it's a little bit Wild West: you have people coming up with new ways of measuring online video viewing but it all seems way off a full solution. We're still talking about a relatively small number of viewers though, so it's not a big issue - yet.”
Here too the TV industry is trying to simplify the challenge through collaboration. The development of initiatives such as Project Canvas in the UK may yet help by shepherding more online viewing through consolidated internet TV platforms. But in the meantime, it's still largely every broadcaster for themselves.
“The thing we're still struggling with is that we are pushing content onto more and more places,” says Marty Roberts, vice-president, sales and marketing at online video management and publishing specialist thePlatform. “With the iPhone and iPad we can't use Flash plug-ins. All the devices are unique, so we have to test all of them. Over the top is a unique challenge too. Some keep your video player. Others have their own.”
His company is directly involved in a potentially useful development announced in May, when thePlatform and the Comcast Media Center (CMC) announced their ability to preserve Nielsen's audio watermarks, enabling measurement of the content produced by eligible telco, cable and satellite operators that is consumed online.
A number of other technology companies with a background in online metrics for advertising are also developing offerings for measuring online content consumption. Many are highly effective, but content providers can end up trying to keep track of information coming in from multiple sources. “We have a customer with six different video analytics plug-ins,” says Roberts. “Each has a solid rationale and they were willing to take the cost to get all those analytics coming out of their player.”
He does think there are some interesting business models emerging in this space. “Quantcast have a really good model: they give their analytics for free to publishers, but take their data and sell it to advertisers. But all of these are all established companies moving into the video space from the TV space or the online display and banner advertising side. There's no clear leader.”
Information sharing is key
A growth in the use of mobile TV will make the situation even more complicated. Mobile TV industry body the bmcoforum is attempting to develop guidelines for mobile TV audience research, but multiple platforms and technologies being used in different markets won't help.
The other fundamental problem is that consumers may only use mobile TV in bite-size chunks, possibly continuing to view the rest of a show on a larger screen later on. “How do you avoid double or even triple counting, if people half-consume content on the phone, then come home to watch the rest?” asks Red Bee's Plunkett. “There's technology that can be used to distinguish those things within the output of a single broadcaster or platform, but unless there is some agreed sharing of that information, it's all spilt apart.”
He does see the fact that much mobile TV broadcasting is based on IP technology as useful, because this means some of the same technologies used to measure online viewing can be adapted for mobile; while the other digital broadcast technologies for mobile TV each have some form of measurement.
“So if one of those became very popular for mobile TV in a given country you would benefit from standardised uniform data. But for now it's a very fragmented space and it's far from clear which mobile TV platforms will exist in a few years' time,” notes Plunkett.
One way or the other, this is an issue the industry needs to settle. And not just for the sake of advertising. “Obviously advertising is important, but broadcasters want to know more about their audiences' interests,” says Plunkett. “If anything, as we move forward into a more internet-connected-TV world, having a better understanding of your audience and mapping interest in the offer you provide is ever more important.”
But is it really possible to envisage a genuinely integrated means of measuring audiences accurately across all of those platforms? “I think so. If you look at the internet, because of the dominance of Google there is some broad demographic data to work with. I'm not sure that level of dominance is healthy, but I'm sure there will be some sort of standardised solution with differentiation at the platform level,” says Plunkett.
Yet even if the technical difficulties can be overcome, service providers, content providers and the ratings and audience measurement providers will still have to find answers to some fundamental commercial questions.
“What is the value of that data and who makes the money?” asks Concurrent's Haddad. “That is a very big nut that needs to be cracked. Nielsen never shared the revenues with content and service providers, but back then the content providers didn't incur costs getting that data out. Today they do and it can cost a lot. So who's going to pay for it?”