Originally Posted by derek500:
“7.7m is the majority of Sky homes, so most.
Sky know the channel has been selected, so if a box is tuned to say Sky Sports Mix there's a good chance it's being watched.”
Firstly, there are 7.7m connected boxes, not connected boxes in 7.7m homes. Homes with multiroom have all their boxes connected, skewing the figures dramatically.
Secondly there are millions of Sky boxes being used by people without Sky subscriptions.
Thirdly, if a box is tuned to Sky Sports Mix, the probability of the television also being switched on is 50%. The television has two states. On or Off. Sky has no way of knowing this, so any judgement it makes on viewing figures from this data is already at least as inaccurate as it is accurate.
The Sky box reports the following for yesterday:
1230 Power on; Channel 103 selected.
1830 Channel 104 selected.
1900 Channel 101 selected.
1947 Channel 0214 selected.
2030 Channel 103 selected.
0215 Box goes into autostandby.
From that information, tell me how many 16-34 men watched The Chase.
The thing is, you can't. It could be any number from 0 to infinity.
That information tells you that someone in the household probably watched Loose Women, someone probably watched Hollyoaks, someone probably watched The One Show and someone probably watched I'm A Celebrity...
It doesn't tell you that they switched off the TV ten minutes in to The One Show because somebody came to the door. Or that someone sat on the remote half an hour later, causing the box to switch to UCB Ireland for 38 minutes until they switched on I'm A Celebrity.
That's why BARB data, with its scientific collection methods, and statistically representative panel, is far more accurate. It's why Sky uses BARB data to sell advertising and sponsorship.
The Sky box data gives Sky some insight into trends over longer time periods, and in particular which programmes are more likely to be timeshifted. But it doesn't tell them how many people are watching to the accuracy of BARB, and never will.