BACKGROUND: GfK is the largest market research institute in Europe. GfK specializes in panel research and surveys in four strategic business units: consumer shopping behavior, consumer media behavior, retail tracking and ad hoc research. The company has annual revenue of $350 million with 4,000 employees throughout the world. GfK Panel Services Benelux has been the leader in consumer behavior research for over 40 years for the Benelux countries (Netherlands, Belgium and Luxembourg).

PLATFORMS: At GfK, we work with several different platforms including IBM mainframes and UNIX systems to collect and process data, Novell and Windows NT network facilities, and DB2, Oracle, Sybase and SQL server platforms. We also work with OLAP tools such as Oracle Express (Web enabled) to deliver monitoring databases over the Internet to our customers.

PROBLEM SOLVED: We saw an opportunity to help our customers find solutions to key business problems. To capitalize on this opportunity, we set up a new data mining service called GfK ConsumerConsult to add value to the market research services we already offered. At the heart of ConsumerConsult is Clementine, the SPSS data mining solution, as well as the commercial and technical expertise of our consultants. The service merges our customers' market research data with other data sources (such as media, retail, sales and marketing data) to develop decision-making models using Clementine. As a result, our customers can now answer difficult questions. For example, where are the buyers and how do they behave? What is the price elasticity for a product? Which products do consumers typically buy in the same shopping trip? What type of promotion is most effective for a particular retailer and manufacturer combination?

PRODUCT FUNCTIONALITY: The ConsumerConsult team uses Clementine to search for causal relationships within data. We convert them into powerful models that, for example, score individuals on their likelihood to purchase a particular product. Then, we give the results to customers in the form of a standalone Clementine Solution Publisher executable. Instead of building their own models in Clementine, our customers simply enter new information into a model built by a ConsumerConsult analyst and get immediate results. We can then easily fine-tune the Clementine model by drawing on our own client service organization and our customers' in-depth market knowledge.

STRENGTHS: Clementine's ease of use is key for ensuring our consultants productively deliver value-added service and communicate results to customers. Additionally, Clementine's powerful predictive models help us address the questions our customers need answered. Clementine Solution Publisher gives us an excellent vehicle to put our customers' knowledge into one complete deliverable.

WEAKNESSES: Some of the data manipulation features are cumbersome, and the graphical user interface could be improved.

SELECTION CRITERIA: We have a long- standing relationship with SPSS MR, the market research group of SPSS. Working with SPSS was a natural choice for adding the data mining capabilities that our customers were demanding. In selecting partners for our value-added services, we wanted organizations like SPSS that are able to service GfK throughout Europe and, in later phases, across the globe.

DELIVERABLES: Our consultants build predictive models by drawing on their data analysis knowledge and expertise and their customers' business knowledge. These models, as well as the data processing steps required for data mining and presenting results, are exported as C code and embedded SQL. The consultants then compile the code and deliver the solution as a standalone executable. Customers run the solution on a new data set or in an ad hoc fashion.

VENDOR SUPPORT: SPSS is in tune with our vision of changing the landscape of market research to increase the value of services we provide our customers. Not only does SPSS share this vision, they also have the tools and personnel that will enable us to succeed in delivering data mining and other value-added solutions. Training in the CRISP-DM de facto standard process model for data mining enables us to communicate data mining in a process-oriented manner.

DOCUMENTATION: The documentation provides technical level information as well as examples for understanding data mining methodology and techniques.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access