Pay TV data delayed til end of year
Research company TNS has won the contract to provide the Australian pay TV industry with an audience measuring service.
The service, to be known as Multiview, will see data collected directly from 10,000 set top boxes. It will measure the viewing activity of Austar and Foxtel households rather than individuals. It is due to begin before the end of the year. Pay TV operators will be keen to have the system in place before the Christmas break. During the summer non-ratings period, the free TV networks generally put out their worse fare, which benefits pay TV which sees larger audiences. This date is later than the original planned launch date of July 1.
Basic data on pay TV viewing habits is already available via Oztam, which also covers free TV. Information reported from Multiview will also include time shift viewing, HD channels, interactive services and video on demand services. It will also measure ad skipping.
The appointment of TNS marks a growing challenge to Nielsen, which dominates the Australian media and advertising industry. Standard Media Index was launched last month with the intention of monitoring media agency spend and at the end of last year Xtreme Information won the contract to provide verification of pay TV advertising logs.
Ian Garland, who will be leading Multiview Analytics, said: “Our partnership marks our commitment to gain new insights into the viewing and behaviour of audiences in subscription TV homes. In addition, the services will provide greater accountability to advertisers across our platforms in areas such as target audiences and interactive advertising.”
Update: Mumbrella did ask TNS whether its appointment had been previously announced as the press release did look vaguely familiar when it arrived this morning. They assured us it had not. Since this they have (very honestly) pointed out that the TNS appointment was made a year ago.
What sort of effect do you think this will have? If the new data can accurately say that x number of people are watching a show on foxtel and it’s completely different from the current ratings one would imagine it would increase advertising revenue for foxtel.
User ID not verified.
What’s the sample size or panel size that OzTam use to measure FTA viewing for 5 City currently? Actually just Sydney will do if you have that…
User ID not verified.
I recall its about 750 homes
User ID not verified.
Thanks Nev.
Would I be reading the reports correctly then if I said *Sydney has an approximate sample size (year 2009) of 1,903 people to represent 4,423,000 people? *Source: http://www.oztam.com.au/pdf/TV.....s_2009.pdf Refer to page 7.
Because that’s less than 0.05%.
Not to mention the fact that ^“All residents and guests register their television usage using a remote control.” ^Source: http://www.oztam.com.au/frequently-asked-questions
Does that mean that when people have the boys over to watch the footy matches, they first sit down and enter in everyone’s age and gender details in the remote control before watching!
He he he. I must have my figures wrong! Surely that can’t be correct?
User ID not verified.
Yes as a general rule the panel is set up with those that live in the household. Guests are registered each time they visit. I cant recall how much detail they give. I think they just register as a guest with limited other information.
I think the australia wide panel is about 2500 homes.
And if i really dig into the memory bank we have a more robust panel that the US as far as Demo’s and actual panel numbers installed.
But at the end of the day its all Voodoo 🙂
User ID not verified.
Coolies – ta for that.
I guess more data we have across the board the better then!
User ID not verified.
Australia wide the panel is 5,050 homes (metro plus regional). There are an average of 2.7 people per household, so we’re talking around 13,500 people a day. However, around 10% a day are excluded for quality reasons (e.g. long periods of continuous viewing to one channel after midnight would be a reason), so we’re effectively talking around 12,000 people.
The procedure is that ‘guests’ are asked to register their age and gender the first time they view the TV that day (that allows for demographic projection) – then they can just record that they were watching the footy using their “temporary” button – all pretty simple stuff. In my day we had 8 buttons on the remote, so there was generally 5 ‘spare’ buttons for guests. In the case of large families there was a 16-button remote so all guests could be registered. Of course there are occasions where not ALL guests can be registered but this is so rare that it really doesn’t matter – and anyway, the audience data would be lower than reported.
Nev, Australia used to have one of the largest (if not the largest) panels for our population size, but countries like the US have recently increased theirs, and I am not sure we can still claim that mantle – but we are still way at the top.
If anyone is interested I strongly suggest you go to the OzTAM website and click on “Television Ratings Guide” because there is a LOT of good information that basically explains everything.
Yes AW, the average persons “represents” around 2,000 people. That is actually a very good sampling fraction. For example, the political polls are generally closer to 1-in-20,000 and they are generally pretty good at forecasting the result (and the margin). Remember you DON’T have to count everyone to get a robust estimate (which is what ratings really are). When you feel crook and the doctor orders a blood test they don’t drain your body dry to make sure they get it right – they take 5ml of your 5.6 litres (let’s say one-thousandth) and use a few drops of that (maybe one ten-thousandth).
I hope this helps to show that among what Nev calls ‘voodoo’ is a hell of a lot of science that at the end of the day produces credible audience data. Remember you don’t have to count everything to get it right – and sometimes when you count everything you get it wrong!
User ID not verified.
From what I’ve been told by people who have looked into it with a background in statistics it’s not a bad method. In some ways though the selection process is flawed as it does not yet allow for time shifting. Everyone I know who can pause live tv does it all the time. They start the programs late and skip the ads. At the moment those viewers are not being counted at all. It will be very interesting to see what happens later this year when they start being counted.
User ID not verified.