Scandinavian satellite broadcasting

The classic problems of measuring penetration and viewing figures

Steen Ulf Jensen

This article looks at the problems of measuring the TV satellite audience in Scandinavia. The national peoplemeters have small samples: 550 households in Norway, 600 in Sweden and 500 in Denmark. This creates problems for smaller broadcasters. With the increase in satellite penetration two measuring problems exist: documentation and control of channel penetration and the measurement of accurate viewing figures. The question of panel and sample sizes is discussed. While the panels are financed by the major channels, data on the smaller channels remain problematic. The need to reduce the cost of producing and installing meter equipment is urgent.

COMMERCIAL television arrived in Scandinavia on New Years Eve 1987 with the launch of the Pan Scandinavian broadcaster TV3. Beaming its signal from London, it broke the legislative monopoly of the public licence-fee, financed state broadcasters in Denmark, Norway and Sweden. Pan-European broadcasters existed of course, but not to the extent that they were able to create a market for commercial air-time or dish and cable penetration. TV3 entered one of the last virgin television markets in Europe. There was one broadcaster in each country, no dish or cable penetration, no cable operators, no independent production companies, no market, no business. Now, seven years later, the commercial television markets are expanding rapidly towards maturity, new channels keep arriving and satellite penetration in Scandinavia is well above 50 per cent (see Exhibit 1) - not as much as in certain central European countries, but still impressive for an area the size of central Europe, and a geography characterised by mountains, lakes and small towns in remote areas. Cabling Norway and Sweden is not a cheap and easy job.


No. of households with TV (000s) and % penetration

For a business like the one described above, there were obviously no peoplemeter systems, since there was no need for commercial documentation. However, for some years now, each country has had national peoplemeter systems. Being countries with small but growing commercial markets and only a few broadcasters means there are limitations on financing peoplemeters. Fairly small peoplemeter panels have therefore been set up: 550 households in Norway, 600 in Sweden and 500 in Denmark. This of course creates some problems for smaller broadcasters. Add to this a continuous increase in satellite penetration and you have two classic and subtle measuring problems:

  • Documentation and control of channel penetration, and
  • Measurement of accurate viewing figures.


It is of utmost importance for emerging channels to have updated and reliable penetration figures. For more mature channels like TV3, which still have growing penetration and sell their air-time on fixed CPMs, the question of penetration is important in that it provides a source of significant increase in revenue.

There are a number of considerations and choices to make: economy versus measurement frequency versus accuracy, for example. In other words there are a lot of compromises to make.

For a channel, the speed of updated penetration figures means millions in revenues. Therefore it is necessary to update penetration figures as quickly as possible, but not faster than natural penetration allows. This means relating the measuring frequency to the growth; slow growth and high frequency will not show anything, with fast growth you can measure frequently and still catch changes. So the first rule is to anticipate growth rates, and determine the necessary measuring frequency based on the estimate. In Denmark where the growth rate has been highest during the last couple of years, questions concerning future acquisition plans for dish and cable connections are regularly asked in order to project future growth. In both Denmark and Norway monthly figures are collected. In Sweden where penetration has grown slowly in the last couple of years, the need has not been so urgent. Here an annual survey is conducted. Perhaps this is not really enough to satisfy the broadcasters' wishes, but on the other hand we are aware of the risk of wasting money on too frequent surveys. Recent events with new satellite positions will create a need for a more frequent mechanism.

Secondly, there is a need, not only to measure, but also to report quickly. This puts pressure on the research company to handle the data rapidly, and also play a role when trying to determine the research methodology. Even though we as television channels want to report and apply penetration figures as rapidly as possible, we serve our own purposes best by having a little patience. To report changes that might be caused by statistical errors is shooting yourself in the foot. A month later, the figure might have decreased due to statistical considerations and you have to adjust it downwards. How do you explain that to the market and still want it to believe the upward changes are real? Therefore some kind of smoothing is needed. In Denmark and Norway we use a three-month running average, thereby creating an unavoidable time lapse, but making the changes more reliable. Until now we have avoided having to decrease penetration figures. So the second rule is a little patience and smoothing, if needed.

The third consideration is the methodology itself. How many to ask, how to ask? There are several problems to tackle. The first is what kind of penetration it is we want to measure: the technical penetration (the channels the household is able to receive technically) or conscious penetration (the channels people think they can receive/are conscious they can receive). Automatically one would say the technical penetration. The only way to measure this really is to check, on the sets, which channels they receive. But is this really the number we want to know? As a channel we do, since it will probably give us the highest figure, but at the same time we do not want it to be used in the peoplemeter panel, because there will be some households which technically can receive without the knowledge of its inhabitants. Do we want these households in our panel? No, not really. They would count as channel households but never watch the channel. Should we then control the panel on both technical and conscious penetration? No, in order to reflect reality - and that is always our purpose - we must use technical penetration and live with the fact that some of 'our' households probably never tune into our channel. So it is conscious penetration we want in the panel, but technical penetration is what we get.

Based on this you cannot rely on which channels people tell you they access. Therefore you have to investigate the sets, and then you have to use face-to-face interviews and that is expensive. But we believe it to be the most accurate methodology. It is our experience that people know very well whether they have primary and secondary channels, but when it comes to supplementary channels they are more vague and therefore produce less reliable figures. So small and/or new channels would opt for face-to-face interviews, but often they are not able or willing to finance this. They have to get the major channels to agree on this. In Denmark we have combined telephone and face-to-face interviews. We conduct face-to-face interviews bi-annually and in addition do monthly telephone interviews. By conducting face--to-face and telephone at the same time we are able to calculate a conversion factor between the two methodologies, and apply this to the running telephone interviews. In this way, the face-to-face survey, which we consider the most accurate, is decisive and the economic approach of telephone interviews takes care of the frequency. So the third rule is conscious penetration, and the fourth, a combination of face-to-face and telephone interviews.


For smaller channels with limited penetration, there is not only the question of determining the penetration, but also inability to control the penetration of each channel in a panel matrix. Most channels must 'flow' in order to limit the panel controls. Their potential audiences can change from day to day and from week to week. One way to handle this problem is to establish a satellite universe as part of the panel matrix, and 'fix' the satellite channels universes by applying a weight as a percentage of the satellite universe. This is done for example in Sweden and Norway. In Denmark Gallup have taken the dominating position of TV3 into consideration and included TV3's penetration instead of satellite as part of the panel matrix, thereby reflecting the fact that the absence or presence of TV3 has a greater effect on the household viewing, than that of satellite channels as such.

Once having established the existence and penetration of a certain channel and somehow controlled it in the panel, the real work starts: how do we establish viewing on the satellite channels?

There are a number of problems to solve, and a lot more to come in the digitalised future. I shall concentrate on a few of the current ones.

First of all there is the question of panel size and the subsequent problems concerning sample sizes. For smaller channels this of course creates constant problems, but they are still weighted against the finance which is necessary for solving the problem, and the obvious limited interest from the major channels to participate in such financing. However the small channels must learn to live with this. Most peoplemeter operators put lower limits on sample sizes. In Scandinavia we typically work with two kinds of lower limitations; a warning, typically with sample sizes between 100 and 50, and blocking with sample sizes below this point, which simply means that the user will not get the opportunity to process data based on these small sample sizes. One can discuss the level of warning and the fact that less than 100 is too small a sample. But in reality, a flexible solution is chosen to satisfy the financiers who, to be honest, tend not to care as long as we can get some kind of data with which to work. But the sample size problem is a problem, and a growing one. As more smaller channels emerge, including national-based niche channels such as Kinnevik's youth channel Z-TV, it will be necessary to solve the cost problem. The need to reduce the cost of producing and installing meter equipment is urgent. And it is not just a question of cutting the cost by 50 per cent, but of moving from order production to mass production, to cut down the margins and multiply the output.

We have had difficulties handing a special problem in Scandinavia, caused by the multiplication of satellites, and the absence of a 'hotbird' position as in other countries, where all dishes are targeted at one satellite position using stationary dishes. In Norway for example we have had two equally watched channels: TV3 on Astra and TV Norge on Intelsat 702. Both channels represent more than 10 per cent of viewing among satellite homes. Each of these satellites are in different positions, and in order to receive both, which most satellite viewers wish to do, they need a motorised dish. This set-up has not only halted satellite penetration in Norway since the necessary equipment is quite expensive: it has also caused measurement problems. When a household has a motorised dish, it has more than one 'channel list'. Again, this forces channel penetration to be measured by physical visits in the homes, since the technicians must measure in advance which satellites and channels on these the households are able to receive. Once the data come in from the households, a combination of channel frequency and dish positioning is used to define at which satellite the dish is pointed, which 'channel list' to use, and which channel is watched on this list.

As broadcasting becomes more digitalised and the number of channels multiply, we shall face more and more problems like the ones mentioned in this article. As broadcasters, we can only hope that the cost of producing peoplemeters declines rapidly, to allow for mass production and bigger panels, and that technology will be able to catch up with the revolution that is unfolding in front of our eyes.


Steen Ulf Jensen

Steen Ulf Jensen

Steen Ulf Jensen is Scandinavian Head of Research for TV3 Broadcasting Group Ltd. In this position he oversees the research activities in the three Scandinavian markets as well as TV3's headquarters outside London. He is co-financier and user of the three Scandinavian peoplemeter systems; the Gallup panel in Denmark, the MMI panel in Norway and the Nielsen panel in Sweden.