Ten chat show The Circle struggles with only 50,000 viewers

given that the pre-noon weekday time-slot is exclusively the domain of a niché audience of unemployed bogans, retirees and stay-at-home mums—the shows’ producers should look further into the figures, to investigate whether it’s meeting the demands for their target demographic. this article doesn’t give concise data, but certainly the age-group of the viewers is far more relevant than simply how large their slice of the (diminishing) advertising pie—surely advertising agencies wouldn’t rely on basic figures such as these.

that said, the shortcomings of their collection methods are growing ever clearer, with the shift toward decentralised, divergent tv sources, the accuracy of that data is rapidly worsening—according to OzTam’s own figures, they assess roughly 0.02% of the FTA viewer’s data; needless to say, a minor shift in one part of their users would create an exponential spike in the national figures. it may be the best/only data available, but it’s high time they considered adopting a new platform(s), e.g. partnership with the TiVo userbase, providing infinitely more precise data.

User ID not verified.

Why would anyone think this poor excuse of “The View” type show work…esp with weak hosts? Did they not learn the lesson from the Nine Network with “The Catch-Up”….

User ID not verified.

Fair comments magic monkey. Mind you, sample size is not everything. Last time I went to the doctor and they took a blood test they took 5ml of blood – he didn’t drain me dry just to be certain.

If you look at TV ratings over time (i.e. multiple episodes) and they keep returning the same sort of number then you can be pretty sure that it’s a fair and reasonable estimate (after all, the TV ratings are estimates and not known truths). In The Circle’s case up until the end of March (Ep 37) it had averaged 57,000 and ranged from a high of 85,000 to a low of 34,000. I think that no matter how much sample size (and $$$) you threw at analysis you’d still end up around the 50,000-60,000 mark … so why waste the money?

However, stay tuned for what Multiview Analytics have planned for Subscription TV data. Mind you, data you get from STBs is NOT viewing data – it is tuning data. The box has no idea how many people are watching the programme tuned on the STB – or indeed if the STB signal is even being shown on TV set that is turned on (many people turn the screen off but not the STB – e.g. my iQ box is always on). To use such data you will still need panel data to establish viewing rates to apply to the tuning data to estimate the total viewing audience, as well as to model the demographic profiles of that audience data.

User ID not verified.

magic monkey mentioned

“the pre-noon weekday time-slot is exclusively the domain of a niché audience of unemployed bogans, retirees and stay-at-home mums”

“target demographic”

and

“it’s high time they considered adopting a new platform(s), e.g. partnership with the TiVo userbase, providing infinitely more precise data”

Methinks the TiVo user demographic may be far removed from the average Aussie therefore would NOT provide “infinitely more precise data”.

User ID not verified.

@John Grono

I feel your analogy with blood is fair — however, just like testing a swimming pool, all the liquid is reciprocally connected so there is an intrinsically similar result no matter which test tube is sampled.
this is markedly different from the way TV ratings are tested and extrapolated — viewers may all be feeding from pretty much the same mass media trough, but in hugely differing environments. the sample of a few hundred people living in the tralier park next to the Windale pokies pub, will be a world away from those enjoying panoramic hilltop views in Rose Bay and Vaucluse, and the difference in viewing habits would reflect this. obviously marketing and TV production departments would prefer to focus on their clients’ individual demographic rather than the aggregated sum of statistical assumption.

there is going to be an issue with observing the viewer to discern their active viewing time for a very long time. paranoia and privacy concerns aside, there are reliably accurate ways to calculate this — for example, based on length of a TV program, and inactive use of the remote, and suspend the data collection during those periods.

@Una

it’s hardly a perfect scenario, but my example of TiVo (or BeyonWiz) as a data collection point is merely for it’s broad-based media sources and the way it’s usage data might be exploited through simple software patches, (this might already be an existing feature used by the manufacturer). in combination with an initial acceptance form and basic user survey, the data would be collected and uploaded via its net connection to the data analysis agency without any effort or additional infrastructure.
ideally this could be a designed feature of all future set top boxes — given that BluRay 2.0 already has a similar web-based capacity, including it in the Freeview phase 2/3 standard is feasible. it would mean going forward, more and more users would be using the system.

User ID not verified.

@ magic monkey. Very fair and well-informed points. As you correctly point out, blood (and swimming pool water) is more homogenous than TV viewing therefore requiring smaller sample sizes. However, You are a tad off beam with the Windale pub comments.

The two samples we have (MTV and RTV) encompass around 5,050 homes. On any given day, some of the data received does “pass edit” (for example, the sort of edits you refer to in your STB comments about inactive TVs and STBs). For example, last week we averaged n=4,651 homes (MTV + RTV – which includea a small amount of overlap in areas like the Gold Coast and Central Coast) covering n=11,929 people. By any standards, that is a pretty large sample – especially given that it is a daily sample.

To ensure that we don’t have the Windale vs Rose Bay scenario the sample is a stratified random sample. This means that the MTV and RTV markets are ‘sliced-up’ both geographically and demographically. That is, the homes are selected to match the ABS census (and inter-censal) population data. The structure of these homes – number of TV sets, household size/lifecycle, STV, region etc – are also strictly controlled to reflect the population. These two procedures ensures that the sample closely matches the population within a percentage point or two.

Of course there is “noise” in the ratings data. In my 20+ years of experience doing this sort of research, I have found that in more cases than not, the variation in the ratings is due to people in the panel changing their viewing habits backwards and forwards frome day-to-day and week-to-week, rather than any deficiency in the sample size or structure.

I also have a quibble with your comment that “TV production departments would prefer to focus on their clients’ individual demographic”. Given that my wife is in TV production and has also worked at the networks (FTA and TV), I can honestly say that I have NEVER heard any call for individual data – TV is a mass medium and it targets broadly and in my opinion always will.

Thanks for the informed and well thought through comments – much appreciated!

User ID not verified.