SPSS - Customers

Halfords Department Store

Situation

Halfords is the leading retailer of car parts, cycles, and accessories in the U.K.

Challenge

As part of an extensive expansion program, Halfords needed to identify the most profitable sites to build its new stores.

Solution

With PASW Modeler's rapid modeling environment for data mining, Halfords achieved much higher accuracy than with previous statistical models. It was also able to quantify more precisely its decision-making process, as well as confidently predict stores' success.

Interested in Halfords Department Store? Download the PDF

Results

Predicted revenue from proposed new store locations with increased accuracy and confidence

Shifting shopping patterns put much at stake for Halfords, a U.K.-based chain of auto supply superstores. As its business followed consumer trends, migrating from downtown sites to suburban superstores, Halfords needed quantifiable information and accurate revenue prediction to help it evaluate potential locations. With a massive expansion program on the drawing table, the ability to confidently select the right new lots was crucial for each new store's future success.

Halfords began with a database containing historic sales figures and site-profile data. Working from 150 randomly selected stores, the company combined two prior years' sales figures with data from a site-quality questionnaire, store-specific data, and census data describing the demographics of each store's coverage area.

To predict store turnover, Halfords developed a data mining model with PASW Modeler. And to assess the accuracy of that model, Halfords tested it on existing store data. The company also compared the software's model with a previously built regression model created by external consultants to attempt to predict store turnover.

The result? When compared to the original regression model, the software's model was much more accurate and easier to understand. Halfords senior management used this model to better understand underlying reasons for differences in store performance.

Very importantly, the software's model was created and maintained by a senior financial manager in the company's property division; he was able to apply his business knowledge, a vital factor in delivering successful data mining solutions. By working with PASW Modeler, Halfords accomplished its goal, and confidently predicted its most profitable new store locations.