BA
In the same way someone with a Level 2 Accountancy qualification couldn't audit the finances of a multi-billion pound company, I don't think you could design a system to interpret the data of 5 million households into anything more meaningful than what the raw data presents.
I have already designed and built a number of data logging devices for TVs so I probably know what I am writing about.
In the same way someone with a Level 2 Accountancy qualification couldn't audit the finances of a multi-billion pound company, I don't think you could design a system to interpret the data of 5 million households into anything more meaningful than what the raw data presents.
IS
In that case, you can answer my question!
I have already designed and built a number of data logging devices for TVs so I probably know what I am writing about.
In that case, you can answer my question!
NG
Should not be too difficult nowadays with the internet of things. Software can process the data in no time.
Technology exists (in a commercially affordable form) for every TV and satellite or cable box to have its own inbuilt data logger recording the times each channel was watched then downloading the data at periodic intervals to a central server for processing and statistical analysis.
Yes - and platform operators like Sky and Virgin / Tivo make that data available or use it internally.
However that data is not that useful for advertisers and broadcasters as it is missing a vital element. It doesn't tell you who, if anyone, is watching. All it records is that the tuner is tuned or that the playback function is running. It can't tell you if there are 20 people watching or none. It can't tell you who is watching either. That information is vital, and is why people meters are used by BARB rather than just dumb set-top box data.
What that data IS useful for is in establishing the metrics of live vs time-shifting. It lets you know what shows are being recorded for later viewing, which shows are being watched live, when the catch-up takes place etc. It doesn't tell you who is watching but it tells you how people are watching. Different data, still of interest, but not to the same degree as overnights and consolidated figures, which have the key demographic data. (Advertisers like to target audience sectors. For that to work they need to know which type of audience watches each show)
noggin
Founding member
Really? Collating all the data from, lets say, 1 million households would take a long long time, weeks in fact. Even if there were daily downloads of data from all those boxes, putting all that information together into a form that we as the public can understand, and that broadcasters can find useful, takes a helluva long time to do. It would require many more people to do that work, and would therefore bump up the price to alarmingly high levels that even big broadcasters would think twice about.
I don't think you've thought about the work involved in this process.
I don't think you've thought about the work involved in this process.
Should not be too difficult nowadays with the internet of things. Software can process the data in no time.
Technology exists (in a commercially affordable form) for every TV and satellite or cable box to have its own inbuilt data logger recording the times each channel was watched then downloading the data at periodic intervals to a central server for processing and statistical analysis.
Yes - and platform operators like Sky and Virgin / Tivo make that data available or use it internally.
However that data is not that useful for advertisers and broadcasters as it is missing a vital element. It doesn't tell you who, if anyone, is watching. All it records is that the tuner is tuned or that the playback function is running. It can't tell you if there are 20 people watching or none. It can't tell you who is watching either. That information is vital, and is why people meters are used by BARB rather than just dumb set-top box data.
What that data IS useful for is in establishing the metrics of live vs time-shifting. It lets you know what shows are being recorded for later viewing, which shows are being watched live, when the catch-up takes place etc. It doesn't tell you who is watching but it tells you how people are watching. Different data, still of interest, but not to the same degree as overnights and consolidated figures, which have the key demographic data. (Advertisers like to target audience sectors. For that to work they need to know which type of audience watches each show)
NG
THIS.
Sure it means that ratings for shows with <50,000 audiences are to be taken with a large pinch of salt, but few people are interested in audience levels this low, apart possibly from Local TV operations and very niche broadcasters.
noggin
Founding member
Just because it can be done doesn't mean it should be done or is cost effective. Indeed anyone with a basic understanding of statistics knows that there is no point collating the data for 60m people when 4000 can give a 99% accurate indication of what people are watching.
THIS.
Sure it means that ratings for shows with <50,000 audiences are to be taken with a large pinch of salt, but few people are interested in audience levels this low, apart possibly from Local TV operations and very niche broadcasters.
NG
noggin
Founding member
I suspect one route that would move us away from people metering is a smart watch system and audio fingerprinting, as has been suggested for RAJAR replacements (but would be equally valid for TV). You wear a watch-type-device, and as long as you are watching TV with the sound up (and not just with subtitles or on headphones) the audio fingerprint of the channel /programme you are watching is matched to the audio fingerprint recorded in your watch.
Not coping with headphones may be an increasing issue though if you want to measure smartphone and tablet viewing.
Not coping with headphones may be an increasing issue though if you want to measure smartphone and tablet viewing.
RI
Exactly. BARB is the wrong tool for the job. I'm interested in pure viewing figures for all TV channels including those outside of the scope of BARB because they are not within the territory of the UK advertising area. Before I ask BARB, have any BARB boxes been fitted to satellite receivers with a dish pointing to a satellite that does not offer any UK channels? If not then what this means is that if, for example, there are a million people who watch these satellite channels in any one day whilst on British soil then there are a million people unaccounted for in TV viewing figures. They may be irrelevant from the perspective of UK advertising but if it's pure viewing figures for all TV channels then it's a significant omission.
Have you got a reasonable accurate figure for how much data a BARB box collects in a day before I ask BARB?
The data logging devices that I have designed and built collect less than 1kB data a day on average per TV as they just record the channel number and the times when the channel was accessed. This means that if 1 million are used then around 1GB of data is collected a day.
Does it really need to go as far as taking the name of the person watching the TV channel?
The data logging devices that I have designed and built were intended for covert surveillance so had no facility to input a name as they were meant to be as invisible to the user as possible. One model was used in a hotel and another model used in a cable TV network in a university.
However non-UK channels are probably of little interest to BARB as they aren't part of the UK advertising area - and BARB is at least partially there to measure UK viewing audiences to set levels of expenditure for UK TV advertising.
Exactly. BARB is the wrong tool for the job. I'm interested in pure viewing figures for all TV channels including those outside of the scope of BARB because they are not within the territory of the UK advertising area. Before I ask BARB, have any BARB boxes been fitted to satellite receivers with a dish pointing to a satellite that does not offer any UK channels? If not then what this means is that if, for example, there are a million people who watch these satellite channels in any one day whilst on British soil then there are a million people unaccounted for in TV viewing figures. They may be irrelevant from the perspective of UK advertising but if it's pure viewing figures for all TV channels then it's a significant omission.
Now you're talking about a sample size of over 12 million boxes, and even more if you were to include Freesat and Freeview. It would make the whole process of gathering data incredibly prohibitive. You're talking about Terabytes worth of data, being transmitted along the internet every night, and being collated in a central area, and it would take so long to actually do, that you'd end up with data backlogs because the area would still be processing from the previous download.
Have you got a reasonable accurate figure for how much data a BARB box collects in a day before I ask BARB?
The data logging devices that I have designed and built collect less than 1kB data a day on average per TV as they just record the channel number and the times when the channel was accessed. This means that if 1 million are used then around 1GB of data is collected a day.
I'm sure it does but how would the box know when its being watched and who's watching?
Does it really need to go as far as taking the name of the person watching the TV channel?
The data logging devices that I have designed and built were intended for covert surveillance so had no facility to input a name as they were meant to be as invisible to the user as possible. One model was used in a hotel and another model used in a cable TV network in a university.
HC
And covert surveillance is useless for what you want BARB to do.
Commercial television sales houses need to know the audience demographic build up for each programme they have to sell the blank bits around and in them.
If you used your 'anonymous and specific numbers watching' you'd never sell specific product advertising in certain programmes.
If you went down that road, you'd end up with toy commercials aimed at children during Countdown, and ones for high backed orthopaedic chairs in X-Factor.
Commercial television sales houses need to know the audience demographic build up for each programme they have to sell the blank bits around and in them.
If you used your 'anonymous and specific numbers watching' you'd never sell specific product advertising in certain programmes.
If you went down that road, you'd end up with toy commercials aimed at children during Countdown, and ones for high backed orthopaedic chairs in X-Factor.
DO
The only information that the people willing to pay for this data collection are interested in is *who* is watching what programmes. The overall totals aren't of much value to them, and neither is the information on viewers watching programmes from outside the territory they broadcast in.
It's no different to web analytics - the page views are of next to no value, it's the information on who is viewing the page that everybody wants to know.
It's no different to web analytics - the page views are of next to no value, it's the information on who is viewing the page that everybody wants to know.
CI
Exactly my point, you're at the basic end of the spectrum. The sort of stuff BARB requires, includes demographic data about the various viewers in the household. What would be less than 1Kb, now gets a lot larger, even as a basic .csv file, because you're not just logging the changes of channels, and the switching off of the set, you're also logging each change of viewer. `Plus, add in to that households can and indeed do have multiple boxes for Freeview and maybe Freesat or Sky or Virgin, so you're talking about multiple boxes in one household, even if that household has just one TV, and a lot have 2 or even 3 TVs. Once you start adding all that together, it gets prohibitive in data terms.
The data logging devices that I have designed and built collect less than 1kB data a day on average per TV as they just record the channel number and the times when the channel was accessed. This means that if 1 million are used then around 1GB of data is collected a day.
Exactly my point, you're at the basic end of the spectrum. The sort of stuff BARB requires, includes demographic data about the various viewers in the household. What would be less than 1Kb, now gets a lot larger, even as a basic .csv file, because you're not just logging the changes of channels, and the switching off of the set, you're also logging each change of viewer. `Plus, add in to that households can and indeed do have multiple boxes for Freeview and maybe Freesat or Sky or Virgin, so you're talking about multiple boxes in one household, even if that household has just one TV, and a lot have 2 or even 3 TVs. Once you start adding all that together, it gets prohibitive in data terms.
IS
Does it really need to go as far as taking the name of the person watching the TV channel?
No, it needs to know their age, gender and all sorts of demographic info. That's what's useful to the owners of channels and ultimately advertisers
The data logging devices that I have designed and built were intended for covert surveillance so had no facility to input a name as they were meant to be as invisible to the user as possible. One model was used in a hotel and another model used in a cable TV network in a university.
Well I'm sure they were very good for what you built them for but they'd be useless for TV ratings!
Just logging how many times a Channel has been 'accessed' is useless,not only do they need to record who is watching but for how long.
Does it really need to go as far as taking the name of the person watching the TV channel?
No, it needs to know their age, gender and all sorts of demographic info. That's what's useful to the owners of channels and ultimately advertisers
Quote:
The data logging devices that I have designed and built were intended for covert surveillance so had no facility to input a name as they were meant to be as invisible to the user as possible. One model was used in a hotel and another model used in a cable TV network in a university.
Well I'm sure they were very good for what you built them for but they'd be useless for TV ratings!
Just logging how many times a Channel has been 'accessed' is useless,not only do they need to record who is watching but for how long.
Last edited by Inspector Sands on 22 March 2016 11:12am
NG
Exactly. BARB is the wrong tool for the job. I'm interested in pure viewing figures for all TV channels including those outside of the scope of BARB because they are not within the territory of the UK advertising area. Before I ask BARB, have any BARB boxes been fitted to satellite receivers with a dish pointing to a satellite that does not offer any UK channels? If not then what this means is that if, for example, there are a million people who watch these satellite channels in any one day whilst on British soil then there are a million people unaccounted for in TV viewing figures. They may be irrelevant from the perspective of UK advertising but if it's pure viewing figures for all TV channels then it's a significant omission.
Who wants to know this information though? They will need to fund the collection of this data, and I'm not convinced there is a large enough market for the stats to make it cost effective?
Have you got a reasonable accurate figure for how much data a BARB box collects in a day before I ask BARB?
The data logging devices that I have designed and built collect less than 1kB data a day on average per TV as they just record the channel number and the times when the channel was accessed. This means that if 1 million are used then around 1GB of data is collected a day.
Do they also record that the TV viewing the output of the set top box was on and switched to the STB receiver's input and not a games console, DVD/Blu-ray player, Apple TV, alternative set-top box etc.?
Do they record that anyone was watching? Do they record how many people were watching? Do they record who was watching? The last bit of data is the only bit of data advertisers are really interested in, and to a slightly lesser degree, broadcasters.
How does your proposed solution work if I leave my Sky box on and tuned to a channel but switch my TV off, and go out shopping, would you still count me as a viewer? What if I leave my Sky box on and tuned to CNN but then watch a DVD, play a game on a PS3 etc.?
How does your proposed solution work with Catch Up iPlayer/ITVHub/All4 etc. on Smart TVs and Sky/Virgin etc,?
Does it really need to go as far as taking the name of the person watching the TV channel?
For the data to be useful it needs to know WHO was watching. How old they are, how much they earn, their gender, their profession etc. That's what BARB collects and anonymises (and viewers with BARB boxes agree to this). Obviously you would need to meet EU/UK data protection standards for any data you collected.
The data logging devices that I have designed and built were intended for covert surveillance so had no facility to input a name as they were meant to be as invisible to the user as possible. One model was used in a hotel and another model used in a cable TV network in a university.
Not very useful then - all you have is a box that can tell you what channel a receiver is tuned to. Not who is watching or that anyone is watching. (You could assume that a remote control action indicates the presence of one person in the room at the time that the remote control is used - but this doesn't tell you much)
I'm not convinced that selling 'tuned channel data' with no demographic or volume metrics attached has a major customer, and thus major funding available to implement.
I suspect the only people who would fund such audience collection are those channels on non-UK platforms that are aimed at audiences in the UK and who don't get BARB ratings for their channels? But if they aren't on platforms aimed at the UK, they are probably quite low-budget (if they wanted to target the UK they'd be on 28.2E and in the Sky/Freesat EPGs) and thus possibly not a great source of revenue? Alternatively they could be platform operators aimed at other countries, who actively DON'T want to know about out-of-territory viewing as if it is significant it could cause them territorial rights issues. (If Polish channels buy content for broadcast to Poland only, finding out you have 500,000 - plucked from thin air - viewers in the UK could be an issue - at least until the EU decides what it wants to do on the subject...)
noggin
Founding member
However non-UK channels are probably of little interest to BARB as they aren't part of the UK advertising area - and BARB is at least partially there to measure UK viewing audiences to set levels of expenditure for UK TV advertising.
Exactly. BARB is the wrong tool for the job. I'm interested in pure viewing figures for all TV channels including those outside of the scope of BARB because they are not within the territory of the UK advertising area. Before I ask BARB, have any BARB boxes been fitted to satellite receivers with a dish pointing to a satellite that does not offer any UK channels? If not then what this means is that if, for example, there are a million people who watch these satellite channels in any one day whilst on British soil then there are a million people unaccounted for in TV viewing figures. They may be irrelevant from the perspective of UK advertising but if it's pure viewing figures for all TV channels then it's a significant omission.
Who wants to know this information though? They will need to fund the collection of this data, and I'm not convinced there is a large enough market for the stats to make it cost effective?
Quote:
Now you're talking about a sample size of over 12 million boxes, and even more if you were to include Freesat and Freeview. It would make the whole process of gathering data incredibly prohibitive. You're talking about Terabytes worth of data, being transmitted along the internet every night, and being collated in a central area, and it would take so long to actually do, that you'd end up with data backlogs because the area would still be processing from the previous download.
Have you got a reasonable accurate figure for how much data a BARB box collects in a day before I ask BARB?
The data logging devices that I have designed and built collect less than 1kB data a day on average per TV as they just record the channel number and the times when the channel was accessed. This means that if 1 million are used then around 1GB of data is collected a day.
Do they also record that the TV viewing the output of the set top box was on and switched to the STB receiver's input and not a games console, DVD/Blu-ray player, Apple TV, alternative set-top box etc.?
Do they record that anyone was watching? Do they record how many people were watching? Do they record who was watching? The last bit of data is the only bit of data advertisers are really interested in, and to a slightly lesser degree, broadcasters.
How does your proposed solution work if I leave my Sky box on and tuned to a channel but switch my TV off, and go out shopping, would you still count me as a viewer? What if I leave my Sky box on and tuned to CNN but then watch a DVD, play a game on a PS3 etc.?
How does your proposed solution work with Catch Up iPlayer/ITVHub/All4 etc. on Smart TVs and Sky/Virgin etc,?
Quote:
I'm sure it does but how would the box know when its being watched and who's watching?
Does it really need to go as far as taking the name of the person watching the TV channel?
For the data to be useful it needs to know WHO was watching. How old they are, how much they earn, their gender, their profession etc. That's what BARB collects and anonymises (and viewers with BARB boxes agree to this). Obviously you would need to meet EU/UK data protection standards for any data you collected.
Quote:
The data logging devices that I have designed and built were intended for covert surveillance so had no facility to input a name as they were meant to be as invisible to the user as possible. One model was used in a hotel and another model used in a cable TV network in a university.
Not very useful then - all you have is a box that can tell you what channel a receiver is tuned to. Not who is watching or that anyone is watching. (You could assume that a remote control action indicates the presence of one person in the room at the time that the remote control is used - but this doesn't tell you much)
I'm not convinced that selling 'tuned channel data' with no demographic or volume metrics attached has a major customer, and thus major funding available to implement.
I suspect the only people who would fund such audience collection are those channels on non-UK platforms that are aimed at audiences in the UK and who don't get BARB ratings for their channels? But if they aren't on platforms aimed at the UK, they are probably quite low-budget (if they wanted to target the UK they'd be on 28.2E and in the Sky/Freesat EPGs) and thus possibly not a great source of revenue? Alternatively they could be platform operators aimed at other countries, who actively DON'T want to know about out-of-territory viewing as if it is significant it could cause them territorial rights issues. (If Polish channels buy content for broadcast to Poland only, finding out you have 500,000 - plucked from thin air - viewers in the UK could be an issue - at least until the EU decides what it wants to do on the subject...)