Revealing …..the “hidden” ratings numbers

Ask any producer or any network exec and they will tell you the same thing: ratings are about more than Overnight numbers.

There is increasing frustration amongst the industry that a show’s performance is perceived by its Preliminary, or “Overnight”, number -that number we are all so desperate to see the next morning.

Headlines are written around them. Shows are branded instant flops if they don’t measure up. Everyone moves on.

But what happens when those Consolidated numbers come in 7 days later? How are those same headlines adjusted and how are audience perceptions changed when we add in the numbers of people who have viewed via their PVR?

What about replays on multichannels and online views? And let’s not even get started on factoring in Regional viewers too -everything revolves around Sydney, Melbourne, Brisbane, Adelaide and Perth.

When any network exec is looking to renew a show, it is the Consolidated and not the Preliminary number that will drive their decision-making.

TV Tonight has taken a look at some of the numbers for 5 shows that have aired recently on the ABC: Gruen Sweat, Gruen Planet, Rake, Adam Hills in Gordon Street Tonight and Doctor Who.

Each of these shows had a replay on ABC2 and were available on iview.

In the case of Doctor Who, it was available on iview before being broadcast on ABC1.

Here you will see how the “overnight” number differs dramatically from the final numbers for Consolidated ABC1 and ABC2. iview Program Plays are similar to page views and represent a ‘click’ on a video. As they do not represent individuals they cannot be directly compared to TV audience data, but they still give us an indication of increased audiences.

All the data below is an average across each series:
Gruen Sweat (4 episodes)
ABC1 Overnight: 853,000

ABC1 Consolidated: 142,000
ABC2 Overnight: 111,000
ABC2 Consolidated: 4,000
Total: 1.11m

iview plays: 56,000
Including Regional & Consolidated: 1.56m plus iview (56,000)

So whilst 853,000 is the number that gets reported the day after Gruen Sweat has aired, it has actually risen to 1.11m by the time Consolidated numbers are in, and even higher with Regional and iview plays.

Gruen Planet (10 episodes)
ABC1 Overnight: 948,000

ABC1 Consolidated: 117,000
ABC2 Overnight: 88,000
ABC2 Consolidated: 5,000
Total: 1.158m

iview plays: 55,000
Including Regional & Consolidated: 1.633m plus iview (55,000)

Rake (8 episodes)
ABC1 Overnight: 800,000

ABC1 Consolidated: 157,000
ABC2 Overnight: 57,000
ABC2 Consolidated: 19,000
Total: 1.033m

iview plays: 83,000
Including Regional & Consolidated: 1.405m plus iview (83,000)

Adam Hills in Gordon Street Tonight (12 episodes)
ABC1 Overnight: 526,000

ABC1 Consolidated: 20,000
ABC2 Overnight: 70,000
ABC2 Consolidated: 1,000
Total: 617,000

iview plays: 14,000
Including Regional & Consolidated: 893,000 plus iview (14,000)

Doctor Who (5 episodes)
ABC1 Overnight: 565,000

ABC1 Consolidated: 171,000
ABC2 Overnight: 82,000
ABC2 Consolidated: 11,000
Total: 829,000

iview plays: 178,000
Including Regional & Consolidated: 1.136m plus iview (178,000)

These examples help us understand audience trends for episodes on other networks too. Shows that are played on Seven, Nine, TEN and SBS may also be subject to replays on multichannels and online platforms.

At TEN’s Full Year Financial Results presentation last month CEO James Warburton said, “In the US advertisers and media agencies are paying more attention to Consolidated ratings numbers and as an industry we feel we don’t do enough to monetise or report on those Consolidated numbers and we plan to tackle that issue here.”

These numbers support that argument.

So next time you see a show has “only” attracted 700,000 viewers last night, don’t forget that’s not the full picture. But the media can’t tell you the full picture because the numbers are not yet available, and by the time they do come in everyone has moved on.

It’s a frustrating dilemma for everybody….

In September OzTAM announced it will lift the number of homes in the survey panel from 3,035 to 3,500 homes and begin tracking online viewing of TV programmes in 10% of homes.

Figures based on OzTAM and RegionalTAM data.


  1. If you just start uploading data from DVRs and websites into the ratings, then it becomes possible for people to falsify the data for their own ends. e.g. you could pay people in China to click on catch up links, via IP spoofing so they appear to be in Australia. The system must have a very high level of integrity for advertisers to trust it. Consolidated viewings don’t matter for the ABC and Foxtel (which is why they publish them and want them known)

    Commercial Networks on the other hand are a different story. Catch-up are worth nothing in advertising, because apart from Go!’s terrible experiment there are no ads in them. At the moment they are run purely to prevent a growth in illegal downloading. DVR viewings are not worth nothing. Some people may watch the ads, some may watch them at 30x and catch the brands names,or watch an interesting ad. And even if you are using a skip button you catch bits of ads, especially the first and last ads in a block. But they are worth much less than a live viewing. Even a live viewing may not be worth that much if people are talking, texting or going to make a cup of tea.

    Due to the recession and changes in viewing habits the network’s revenues have halved. And the technological change here is still nothing like what has happened in the US where TiVo has been around for over a decade and Hulu, Netflix and iTunes all show TV episodes.

  2. @Secret Squirrel

    Yes, 2000 is reasonably accurate for total viewers. And you need to make large increases in sample size to deliver small increases in accuracy beyond that point.

    This of course assumes that the sample is representative of the population. They try to ensure this is the case but when Oztam switched to the current system there was significant change in viewing figures. There was quite a difference between the two samples, even though they were both carefully selected to be representative of the population.

    The larger sample size of 3500 increases the accuracy when you start drilling down into demographics.

  3. Interesting article David. I’ll admit I don’t let the ratings decide what I want to watch obviously. I also like and appreciate the replays on secondary or primary channels. Sometimes it’s what I rely on to see a particular show. If they plan it well it could sometimes convert to prime time viewing. An old example is Fringe. I used to watch the late night repeats of the first season on Ch 99 and for the second I watched the prime time shows. So they are important.

  4. cynical old codger

    @David Knox 10-43am 6-12-12…..More boxes Per Capita than US?

    Does that included/exclude monitoring by the cable companies connections, that have been a major part of TV scene for just about forever in the USA, and I would have thought they would only miss monitoring the illeagal connections?

    I have never had cable, but I always thought that free to air TV, came with any cable connection, nor do I have any idea about Satelite in regional areas, only that many areas have digital switchover problems, and get everything but their local channels.

    Could Big Brother know more than they are prepared to let on.?

  5. Secret Squirrel

    @Guy – in that case you should understand the difference between estimation error and margin of error, and also how to calculate standard errors and confidence levels. These would then tell you that a sample size of 3,500 is plenty.

    If you can come up with a business case for a technically feasible way to somehow collect, store, and process significantly more data, or a better way of doing that with the current amount of data then go and make yourself a million dollars.

  6. One of the most interesting articles ive read for a while. good job david.
    My question/issue is to the marketing exec’s. How do you justify to your Boards spending millions of $$ (in some cases) on such a questionable measurement system? 3,000 boxes australia wide = wow.

  7. @Secret Squirrel

    I know what i am saying and i know my research. What i am saying is that these numbers are shocking and considering all this technology we have now, the way ratings are collected could be done a lot more accurately and a lot better than they are now, not just here but all over the world.

  8. Anyone who has studied statistics at a university level would know that 3,500 is enough data for a accurate analysis.

    On the ratings, most people see it as a way to track what shows are good and not, which shows to avoid. Not everyone sees it from the eye of the advertisers the information is actually aimed at. To the masses, Total people in cities is what immediately catches the eye, not demographics.

  9. Yes i know, but you really do, do a brilliant job, either giving us Breaking News, Interviews with Network Staff and really interesting Articles and much more.

  10. Ok! Sorry David. Yes! well done on this article. You do a brilliant job, and you are indeed a very hard worker. Well Done!!! And that’s why your website, this one, TV Tonight is my and everyone else’s best website in Australia and possibly the world, because to me, it really is the Best!! In the World!!! You deserve that Award!! Well Done! Again David Knox. A true Legend in Australia!!!!!

  11. Thank you David once again for another informative article!
    Love your work and wish you a great break once you go on hiatus (and I reduce my consumption of the best TV website while I too have a break soon).

  12. Secret Squirrel

    (sorry, last one)

    @Mac – PVRs, etc have no way to collect that information. OzTAM would still have to supply a ratings box for each TV in each ratings household.

    Of course, you could imagine that the govt could force manufacturers to build this functionality into those devices, pushing prices up, but you would still need to connect them to the internet, and then build a back-end capable of receiving the daily logging records from millions of boxes, storing it all, and collating and processing it in time for the morning release. It ain’t gonna happen.

  13. Secret Squirrel

    @Guy – it’s actually about 14m across the 5 cap cities. The statistical methods used are based on years of research and well supported.

    While it is possible that there are a whole heap of people in advertiser-friendly demos who watch a program that almost nobody with a ratings box does, it is extremely unlikely. If the sampling is done correctly then, at the current sample size, doubling the number of people will only provide a tiny increase in accuracy – it’s not worth the additional cost.

    Their is ample info about this on the net if you are genuinely interested in improving your understanding.

  14. Secret Squirrel

    @craigj77 – program viewing information is recorded minute by minute and any advertiser worth their salt uses this data, not the aggregated numbers we get to see for free.

    If you change channels during an ad break, that will be recorded and reported. If you just skip the ads then that won’t count. Of course they can’t tell if you mute the sound and jump on the net / check your messages / read a book / play with the cat but it’s the best measure that they have right now.

  15. @David

    I did know that which begs the question why the US has even less then we do. I guess thats how things are going to be. I do think some shows don’t deserve the dismal ratings they get but its how this industry is so until they change it thats how it will be.

    I mean in the day and age we live in i don’t understand why we can’t plug in the LAN into my PVR or MyStar box and send the data from shows i watch to OzTam. I think this is the way the ratings have to go in the future to get a more broader audience scope.

    • Sorry I am not chasing up other shows, it was enough research to do this. The point of the article is to demonstrate how all shows attract additional eyeballs. iview numbers are sourced from WebTrends data.

      Mac: Data is available, but Consolidated data takes up to 7 days to report.

  16. The Iview lisitng is that strictly through the ABC iview website or does it include the smart Tv streaming ? A lot of people I know are heavily using their smart tvs to watch catchup tv

  17. It’s not so much the 3500 households, it’s the same 3500 households, for years. So if you’re a rusted on Home and Away fan, you’re never going to check out anything else which is on at that time. Yet the people next door may watch a variety of programs at that time.

    So that one household is supposed to represent 1000s of other households around the country.

    Yeah, right.

  18. Doubt it! But the consolidated figures always paint a better picture for TEN.

    The past 3 weeks Homeland has been the most timeshifted program. Audience up 20-30%

    So 650,000 up to 850,00+

  19. I find it amazing that 3500 households determine ratings for 21m people. Seriously the survey panel should be much greater if you ask me in all countries not just this one but i suppose that will never happen in the near future and good shows are being cancelled because a select bunch of people doesn’t watch them.

    I know they are selected for a certain reason but still some shows may have still be on the air if there was a wider survey panel both here, in the UK, in the US.

  20. While this is all well and good for the ABC, I imagine selling advertising based on these figures for the commercial networks is very difficult. Half the reason I IQ stuff and watch it back is to avoid the ads.

  21. This is very interesting.

    But do commercial networks sell advertising based on overnight ratings or 7 day consolidated.

    I know they don’t sell advertising on online catch ups (that is sold seperately)

    If the networks don’t sell advertising based on 7 day consolidated viewing – then they aren’t getting paid for those extra viewers.

Leave a Reply

You must be logged in to post a comment.