Classic Paper: ‘An approach to fusing market research with database marketing’, Barry Leventhal, 1997
16 July 2013

In this era of ‘Big Data’, one challenge is the integration of survey data with other forms of data, and to do so in a way that does not break the MRS Code of Conduct, and indeed the law, on preserving anonymity of participants. 

The issue first emerged in the mid-1990s as organisations discovered that their transaction based customer databases provided limited details of their customers, and sought ways to enhance this limited information with data from other sources. 

The first step had been the development of geodemographic systems in the late 1970s, firstly by CACI (Acorn) and then CCN (Mosaic), leading to other products which had transformed the information available at postcode level. 

But this was never going to be enough. The first published paper in the field of integrating survey data with customer databases that I’m aware of was presented at the 1997 MRS Conference, in a session I chaired, referenced by Leventhal, describing the TNS/Claritas method. 

Dun Humby linked with BMRB, using the TGI survey to create a product called First T, which I trialled whilst still at the AA in 1999. 

However, Leventhal’s paper - published in 1997 - provides a very detailed methodological description of how the FRUITS product was developed by Berry Consulting using data from the then NOP FRS financial services survey.

Leventhal not only describes the methodology, but also includes two short case studies describing applications. Whilst these products were methodologically robust, and ethical in terms of preserving anonymity of participants, they didn’t match expectations in terms of popularity. 

Maybe this was because the process led to the development of models, or segmentations, rather than delivering customised data on individual customers – never really possible with the small size of the survey data inputs, plus the confidentiality issue. 

For example, even though First T used three years of TGI data, at most 75,000 records, so an organisation needed a large database, in the millions, to secure sufficient matches to create reliable models. 

Obviously, the volume of data available today, and the number of sources available to many organisations, is much more complex than the situation in the 1990s, but the principles described in the paper still apply today – the need for a robust methodology; the issues of data privacy and anonymity, and, a high degree of transparency if potential users are to trust marketing data integration products.

Read this classic paper now

‘An approach to fusing market research with database marketing’, Barry Leventhal, first published IJMR Vol. 39 No. 4, October 1997.