SPSS - Customers

Herlitz AG

Situation

Herlitz AG, a leading German manufacturer of office products, is offering its trade partners the information that has already helped local corner shops increase their sales: location-matched product lines.

Challenge

In the past, a grocer knew from experience what his customers wanted. He understood which products would sell best and how to display them. And he did this with great success—sales increased, cash registers pinged. Today's supermarket branches have lost sight of this knowledge amid a mass of customers and products.

Solution

Some large chains are already using data mining to help them provide the best range of products to suit a specific location and discover the hidden preferences of their customers. This is where Herlitz AG is breaking new ground.

Interested in Herlitz AG? Download the PDF

Results

Using the information gained about products, locations, and specific regional characteristics, the company is helping its partners to specifically analyze the buying habits of their respective customers and find the best location for a specific item.

Optimizing the product location ultimately benefits the supplier, as well as the consumer, since the company can respond to the specific needs of its customers and, at the same time, carry fewer expensive, slow-moving items. In addition, customers get a clear overview of the products offered and are able to access them easily.

Herlitz AG is one of the leading manufacturers of office products in Europe and has more than 20 years of experience in space and category management. Based on the individual requirements of its European trade partners, the company offers its customers tailored solutions for optimizing their product ranges, particularly in self-service markets.

Herlitz AG's aim is to to increase sales using its comprehensive knowledge of customer requirements and the factors that influence sales. The company derives this knowledge by analyzing the reputation and range of products available, as well as regional differences. The main question here is: Which factors result in one item being sold more than another?

Herlitz AG first used data mining software to analyze these factors. Its initial project, started in October 1998, provided the first insight into how to place products in the best locations. However, in view of the constant changes in the factors that influence consumers' buying habits, it quickly became clear that such an analysis would have to be automated and performed on a regular basis.

Herlitz AG was seeking data mining software which, in terms of automation, could reproduce, structure, and optimize the entire data stream. At the same time, the software had to be user-friendly, intuitive, and economical. Ultimately, the decision was made to select PASW Modeler. The decision was made based on the software's user-friendly interface and its ability to analyze large amounts of data in one step, and export these analysis results into another system for further processing. Another plus point was the fact that Herlitz AG had already used SPSS Inc.'s solutions for analyses in the past.

A meaningful analysis can only be as good as the data it is based on. For this reason, the first step Herlitz AG took to provide a range of products tailored to a specific location was to focus on preparing the data. This involved cleaning up data from the internal data warehouse and enhancing this data with third-party data. All sales data for each item were compiled from the current range of products available at all points of sale. Information about the location of a point of sale was added to the existing data to help analyze not only the internal sales data, but also information about a region, social and demographic structures, the competitive environment, and particular regional characteristics. Rather than reverting to statistical samples, all the items, including all points of sale, were analyzed individually. The datasets comprised an average of 4,000 cases with approximately 30 variables. The data was then transferred into PASW Modeler from a Microsoft® Excel® or ASCII file format.

Selecting and enhancing the data with external data formed the basis for the next step: Creating a process flow diagram in PASW Modeler to classify sales into below-average and above-average categories. The team chose a decision tree process for the analysis, which was automated using the PASW Modeler language.

On the basis of these results, at least one rule or profile indicating when sales are above or below average was generated for 68 percent, or 4,420, of the 6,441 items analyzed. Using a tool developed by Herlitz AG, a range of products was then specified for each individual subsidiary which, taking into account all the special features of the location in question, would guarantee improved profitability.