IPA TouchPoints Integration

The IPA TouchPoints Hub Survey launched earlier this year provides insight into cross-media consumption, with the ability to analyse this by a wealth of demographic, behavioural and attitudinal information. Roger Gane, research director at the RSMB and one of the UK’s leading audience researchers, looks forward to the next stage of the survey…

There are inevitably limits to the ‘granularity’ of the media behaviour TouchPoints can offer; moreover the survey cannot itself reproduce the ‘currency’ audience levels delivered by the industry audience services. This is where the next stage of TouchPoints – media data integration – comes in.

RSMB was commissioned to undertake this project.The industry media currencies to be included were: BARB (television), NRS (national newspapers and magazines), RAJAR (radio), JICREG (regional newspapers), POSTAR (outdoor) and CAA (cinema).

To further enhance the planning capabilities of TouchPoints, it was also agreed to include KMR’s TGI, thus adding product information.

There were a number of challenges to be confronted in achieving all of the above. The first issue is that of the Hub Survey itself. Its availability was a considerable potential benefit in terms of the quality of the integration (as opposed to simply using demographics common to the various currencies). But the sample is relatively small, relative to those of the media currencies, and it needs to be weighted to universe profiles derived from a large high quality random probability sample.

RSMB’s solution was to fuse the Hub onto the BARB Establishment Survey (ES), which provides a sample of 50,000 adults. For each TouchPoints respondent, a larger number of ES respondents with the same demographic profile were available. The result is a re-engineered Hub Survey representative of all geographies and planning groups.

The integration process adopted varied according to the nature of the data available from the various industry surveys. The currencies for TV, radio and national print media were integrated using respondent level fusion from BARB, RAJAR and the NRS. The principle was to match in the media currency respondent (the ‘donor’) with a re-engineered Hub Survey respondent (the ‘recipient’) in terms of demographics and ‘media imperative’ profiles.

The ‘media imperatives’ were derived from the Hub, for instance a heavy TV viewer, and of course it made sense to fuse this with a BARB heavy TV viewer; so it can be seen that the availability of the Hub should improve on the quality of the fusion achievable using demographics alone. TGI data have been similarly integrated with TouchPoints using fusion.

Integration for outdoor advertising and cinema was achieved by calibration. That is to say poster exposures and cinema visits as measured on the Hub were calibrated to the levels established by the POSTAR and CAA currencies. Regional newspapers were integrated by profile matching the JICREG currency data (which is produced in aggregated form, rather than at respondent level) to the Hub.

Whilst the currencies discussed above cover a high proportion of British display advertising spend, they are not entirely comprehensive. It was decided that three further communications channels needed to be included: the internet, direct mail and SMS commercial messages

Although no industry currency exists for these, the Hub Survey included a range of questions on each of them. Treatments have been agreed for each of these to allow them to be included in the integrated database.

So we now have a comprehensive integrated dataset covering nine media, TGI product information, and TouchPoints Hub Survey behavioural, attitudinal and demographic data. However, the main application of this integrated database will be multi-media schedule reach and frequency analysis. Apart from TV, the media currencies provide only short-term exposure measures and depend upon varying expansion models to estimate longer-term contacts with an advertising schedule.

A common denominator was required and RSMB’s solution was to adopt an approach based on personal probabilities. I will not go into the detailed approach; in summary it is based on a segmentation analysis of each currency to define homogeneous groups within which a binomial expansion is used to estimate each person’s probability of making one or more contacts with a particular schedule. The approach has been tested to ensure that duplications between titles, channels etc., are preserved.

The completed database has now been provided to a number of bureaus who are developing systems to handle it – and in particular the multi-media frequency distribution calculation required to run the schedules. This is a major task and the bureaus’ role in this should be acknowledged.

To conclude, I would like to remind you of the purpose of TouchPoints, by quoting the IPA’s Lynne Robinson (interviewed in Research World, July/August 2006).

“TouchPoints is primarily about communication planning… to provide a gateway across media currencies and other data sources but not to act as an alternative to current industry research.”