Dominik Kaiser: "We are not the problem".

WW: Dominik Kaiser, you are not the only one who is unhappy with the new TV data. But only you went to court, only you are constantly in the media. Seen in that light, you are a lonely fighter ...
Dominik Kaiser: The impression is deceptive. Telesuisse also communicates effectively with the media; ten regional TV stations have just terminated their contract with Mediapulse because they have no confidence in the correctness of the figures. But it's true: Goldbach, the marketer of the German private stations in Switzerland and also of us, works more in the background. We, too, waited a long time before voicing our demands publicly, because basically the ratings problem doesn't belong in the media, but would have to be solved differently. Until the end of March, we tried to find an internal solution with Mediapulse. Only when this did not work, and because we received no answers to our letters and demands, did we go to court. Immediately after the high court granted the super provisional injunction, we offered Mediapulse to find a solution without publicity. Mediapulse decided on a different path and went to the media. Since then, we have also been communicating openly about the ratings debacle.

As mentioned, Goldbach is keeping a very low profile. As an observer, you could get the impression that 3+ is being sent ahead and letting you fight the feud.
That is your interpretation. Goldbach is very active in the background. Together with Publisuisse and Mediapulse, the company has just drawn up an action plan with around 50 points. This includes almost all the points that we are also demanding. But to answer your question: It takes courage and perseverance to stick your neck out. However, based on our work with Mediapulse and Publica Data so far, I believe that this is the only right way to go.

What do you mean by that?
Cooperation in the past was often not constructive. Even in the old system, we noticed and questioned irregularities in the data. In very few cases were our inquiries answered completely, and when they were, it was always only after repeated inquiries. Another example: We have been trying for eight years to obtain from Mediapulse, in return for payment, the pin data that is customary internationally and was also available in the old system - to date without success.

But the changeover of the measurement system involved the industry. The Board of Directors and the User Commission, both made up of stakeholders, waved the change through. The discussions didn't start until the data was in.
The discussion started after Mediapulse published the data in January and immediately blocked it again and it became clear that something could not be right. We already demanded a test of the whole system at the end of January. The User Commission (UC) was not involved in many decisions. As far as I know, only one assignment rule was discussed in detail in the UC, for example. The others were not discussed at all or were addressed only very superficially. Assignment rules are critical in the new audiomatching system because the system does not know directly what is being watched, as it used to, but often has no clear information. To date, after much pressure, the UC is only aware of a small portion of the assignment rules. If committees are not informed, they cannot have an active say. That the UC has signed off on everything is therefore not true.

A delicate point. Can you substantiate this?
Since 3+ did not have a seat in the UC until now, I do not have any minutes. We have only been allowed to have a representative since May 13. I got my information from conversations with commission members.

Nevertheless, shouldn't the broadcasters and marketers be taking themselves by the nose? They could have reacted before the data was available.
I can only speak for 3+. We insisted too little last year when Mediapulse informed. We asked whether the panel households could be recruited representatively in the short time and connected on time, whether the different feeds of the German stations could be mapped with audiomatching, or whether a feature film running on three different stations at the same time could be assigned to the right channel. The answer was always: no problem. The rules according to which the signals are assigned were not explained to us. We would have had to ask and insist. The fact that we didn't do that last year is a big mistake in retrospect. Especially with the example of the movie or series that runs on several stations and is played back without advertising, it is technically impossible, with the audio signal alone, to recognize on which station it was recorded. The system has to guess. Besides, we begged in vain until the end that there should be at least one month of parallel operation.

The aforementioned problem of allocation could be circumvented with watermarking.
That's why we already lobbied for watermarking last year. But Mediapulse assured us that this was not necessary and that watermarking would only be introduced later, if at all. We should have exerted more pressure here, too.

You always emphasize that 3+ is interested in a constructive dialog. What is your proposed solution?
The industry needs numbers - not only our customers and the agencies, but also ourselves. As long as we don't have data, we can't optimize our program planning and it's difficult to decide which shows to buy or produce. That's why we are very interested in a quick solution. That's why we made concrete demands for the first time at the end of January on how to restore the credibility of the system, followed by a major effort to find errors, and finally compiled a catalog of points that would have to be adjusted. Basically, there are three levels involved: a. the panel corresponds to the Swiss population and is therefore representative, b. is the system measured correctly and does it work, and c. does the industry know the background and details of the TV research, in other words, is the system transparent. On all levels, there is great need for adjustment or catching up.

What are you doing to get the numbers back as quickly as possible?
We are ready to actively collaborate. But Mediapulse seemed little interested until recently. Even Mediapulse has not disputed so far that the distribution types as well as the urban-rural distribution are not correctly weighted and that the PC-only households (households that do not have TV) can be mapped as we suggested. These undisputed errors must be corrected retroactively. They lead to a noticeable distortion of the audience figures on average and to considerable differences at the level of commercial breaks and programs.

Can you elaborate on this?
If we compare the panel's figures with those of the Federal Statistical Office, for example, the rural population is heavily overrepresented. The proportions of the distribution types (IPTV, cable digital and analog, etc.) would have to be adjusted to the information provided by the providers. And a third point: As long as TV use on tablets and smartphones cannot be recorded, households without TVs must not be included. We know from online TV providers such as Zattoo, Teleboy and Wilmaa that the vast majority of web TV usage is via tablets and smartphones. Particularly in the case of PC-only households, TV usage is massively underestimated if tablets and smartphones are not included.

However, this is not just a methodological problem, but is related to Apple, whose policies do not allow apps to communicate with each other.
That's true, and without it, it's not possible to measure PC-only households in a representative way. That's why we shouldn't even pretend that this is the case. The fact that tablets and smartphones are not measured has only been known since the beginning of winter. If this had been known earlier, the decision would certainly have been different.

Are these all the demands?
We also propose to test the system on the fly and over the raw data. Switzerland is probably the most complex country for audio matching. The signal has to be compared with almost 400 TV stations and many more radio signals. The two-channel sound leads to a doubling for some stations, stations with country-specific advertising windows like RTL have three feeds (Switzerland, Germany, Austria). In addition, many programs run more or less simultaneously on different stations. The test would show whether the system assigns signals correctly even in difficult situations. Another demand: transparency. We want the research basis to be disclosed. We want to know the assignment and weighting rules and have insight into the New Establishment Survey to understand the basis of the panel. This is in line with the EBU's international standard GGTAM (General Guidelines for Television Audience Measurement).

Read the rest of the interview subscribers online or in the print edition.

Interview: Isabel Imper and Pierre C. Meier

Federal government takes Mediapulse measurement system under protection