The Future is Hybrid

Benjamin Franklin famously said that: “Nothing can be said to be certain except death and taxes” But in our own world of market research, surely ‘declining response rates’ are an equal certainty. The trend has been with us for a very long time.

Face-to-face interviewing has almost disappeared in some countries – ESOMAR reports it as accounting for just 8% of MR spending in 2014, compared to 24% in 2004. Online and digital data collection has been the big winner, growing from 13% to around half in 2014.

Lower response rates generally mean that research results are less representative – although steps can be taken to minimise bias through careful sample selection and weighting. But response rates overall can be improved by offering people a choice of methods for communicating their answers to our questions.

A second long-term trend in our industry has been the tendency to try to fit more and more questions onto a study or to demand more and more of a participant generally, as our clients seek to understand their customers as deeply as possible.

Some studies still ask people to spend two hours or more telling researchers about the products they buy, the media they use and the lifestyles they lead. It is detail that clients want and ask for.

In the world of audience measurement, we now want to know not just when people are watching TV or reading a newspaper, but which platforms and devices they use, whether they look at ads and respond to them, whether different environments are more conducive to them being receptive to advertising and so on.

It is a lot to ask and can be both difficult for people to accurately recall such detail or for them to be willing to participate in such an exercise.

Television viewing has long been metered in most countries, meaning people don’t have to remember everything they watch and record it somewhere. But it has been limited to viewing inside the household and restricted to monitoring viewing on television sets. An increasing amount of viewing now takes place on other devices and outside of the home, which the old generation of peoplemeters is struggling to keep up with.

Work is now going on in many countries to combine data on TV programme streaming (e.g. from ‘catch-up’ players or other services) and on out of home viewing with the traditional TV set viewing in the household.

In the area of newspapers and magazines, the need to measure reading across multiple platforms has become urgent, as reading of printed publications declines. The nature of news is such that people may glance at their Smartphones or tablets several times a day in various locations to read headlines or longer articles. It will be hard for them to recall each occasion or, if they are browsing news aggregator sites for example, which publisher brands they have accessed. Yet this is what we need to know.

If we can’t ask them – as we have long done for printed publications – we need to capture this information in some other way. In around 20 countries, panels set up to track online and/or mobile usage have allowed their data on visits to publisher sites to be ‘fused’ onto readership surveys. This allows users to estimate readership across and between the different platforms.

Important components of any successful fusion are the ‘hook’ questions designed to help us find similar people within a survey sample and the separate internet or mobile panels. We match them on obvious things like gender and the region in which they live, but can also use other information such as ownership and usage of certain devices.

Another way of addressing the challenge of over-long surveys is to use ascription. By asking a set of core questions to an entire sample, but then splitting the sample into two or more matching sets and asking different questions of each set, answers can then be ‘ascribed’ or imputed from the set asked the questions to the matching set of people not asked (who, in turn, have the answers they gave ascribed to the other sets). In this way, we can generate a database far larger than a single questionnaire would allow.

There is no doubt that audience measurement is moving in the direction of supplementing traditional survey research with other data, either captured in separate surveys or in other ways.

A good example of the latter is Out of Home measurement such as the Route study conducted by Ipsos in the UK. In this, a panel of people are asked to carry a GPS-enabled device with them over a period of two weeks to capture where they travel. The data captured is then combined with a host of other information including road and pedestrian traffic counts, information about the poster locations themselves and where they can be seen from and likely ‘visibility’ to passing cars and pedestrians.

The goal, in the end, is to provide accurate estimates of the potential audience to advertising on every advertising frame.

• Read more on our audience measurement and understanding expertise
• We welcome your comments below