It sounds like the 21st Century equivalent of counting the number of angels on a pin-head. How do you measure TV audiences when viewers have a thousand digital channels to choose from, and the existing measurement techniques won’t work?
Without audience figures, advertising-funded channels are doomed. Although pay-per-view and video-on-demand will provide their own audience figures – with each viewing occasion turned into a transaction, recorded like a phone call – the need for BARB-type figures will be essential well into the next millennium.
ITV will face yet more competition for advertising, not just from Chan-nel 4, Channel 5, BSkyB and the existing satellite channels, but from dozens of digital channels – whether beamed from space, via cable or conventional aerials. The BBC, becoming ever more accountable, also needs reliable ratings to support its claim to a universal licence fee.
This week, researchers from all over the world are meeting in San Francisco at an Esomar seminar to compare notes on the problem. But already it seems two distinct schools are developing, those who believe the answer lies in putting a coded signal on every programme, to be read by the peoplemeters, and those who don’t.
The coding camp is led by a distinguished American researcher, Gale Metzger of SRI, backed by the combined forces of the mighty US networks. He is working on a pilot in Philadelphia and will be revealing his latest progress in a paper in San Francisco.
Leading the non-coding school, is our own Taylor Nelson AGB, contractors to BARB until the year 2000, and anxious to hang onto the business thereafter. Last week, it showed a handful of agencies and TV companies the latest development in the race to crack the digital problem. It’s a technique called Picture- Matching, which enables the peoplemeter to work out what is being watched at any time, simply by analysing the picture on the screen.
This struck me as a most unlikely solution to the problem. With more than 30 channels at home, it often takes me minutes to work out which one I’m watching, even when they put a little logo in the corner. When there are hundreds of stations, the chances of working out which one is which, second by second, purely by looking at the image on the screen, must surely rival that of winning the Lottery jackpot.
The men from AGB didn’t tell us the odds, they just showed us they can do it. But first they explained why they don’t simply transfer the existing BARB system to digital, and why they don’t believe coding will work.
“The existing analogue peoplemeters measure the frequency each channel is broadcast on,” says AGB managing director Mike Kirkham. “The problem with digital is that it transmits multiple pictures and sound signals, and a single frequency may contain up to seven different channels.”
So why not get each channel to put a code on its signal, as favoured by Metzger? “It’s easier said than done,” says Kirkham. “It requires the co-operation of the set-top box manufacturers, and of every broadcaster, who must put the code on everything they put out. That will be complicated and if they won’t co-operate, or put the wrong code on, the system won’t give an accurate viewing measure.”
He says the advantage of PictureMatching is that researchers can measure viewing whether the broadcasters co-operate or not. But how is it done?
Personal computers linked to the meters take continuous snapshots of parts of the TV picture, measuring their brightness and comparing them with a master file of all transmitted channels. I watched it measure six channel changes within two minutes and get them right to the second. But that was with just four channels – how will it manage the hundreds offered by digital?
AGB says a desktop PC can handle 32 channels at present and the software can cope with a thousand channels. “Four or five years ago this couldn’t have been done,” says Kirkham. “It’s the enormous increase in computing power that has made it possible, and of course this increase will continue.”
It can even distinguish between channels that are transmitting the same pictures. AGB was experimenting with the system in ten homes during the last World Cup final, which was carried live by both BBC and ITV from the same picture feed. By tracking backwards or forwards to a commercial break, or an on-screen graphic that would identify the channel, it coped with virtually all channel changes.
PictureMatching meters are being installed in a Taylor Nelson AGB panel of homes in the Meridian area, and though digital TV is still some way off, Kirkham says it already offers benefits in measuring cable TV. The problem for AGB and BARB is that each cable franchise has its own line-up of between 40 and 60 channels, and they are often changed at short notice. BARB has to be notified of each one in advance.
“On the night of the Bruno-Tyson fight we had to face 93 line-up changes across 59 systems,” says Kirkham. “PictureMatching would solve this problem too.”
Researchers in British agencies and TV companies are full of praise, and grateful that when digital TV comes, they can be confident it can be measured. However, with the resources being pumped into Metzger’s Philadelphia laboratory, AGB may still have to convince people that coding is not the way forward.