Fortis Bank marketing campaigns gain in efficiency

SAS^® Enterprise Miner™ reveals hidden patterns

Fortis Bank is the largest financial institution in Belgium. However, size simply is not enough to guarantee success anymore. If a bank is to remain competitive and make sound strategic decisions, then accurately identifying customer reaction to marketing campaigns is absolutely essential. Fortis owned a rich – but relatively untapped – customer database. The bank's recent adoption of data mining technology has allowed them to substantially increase the successful hit ratio of marketing campaigns.

Which customers are most likely to purchase a new investment product? How will the market react to a drop in interest rates? Which client profile shows the highest risk of not being able to pay back a loan? These are just a few of the questions Fortis Bank faces every day, and perfect examples of the issues to which SAS data mining technology provides comprehensive answers.

Making the most of hidden data
"Given the large number of customers at Fortis Bank, we obviously work with a huge amount of data. For both efficiency and budget reasons, we cannot direct our marketing initiatives to every single one of our customers," says Joseph de Brouwer, Commercial Analysis Manager at Fortis Bank. "That is why we now rely on SAS technology. With this tool, our marketing analysts rapidly obtain precise lists of the profiles they need for designing a new campaign." As a result, not only have the bank's marketing expenses gone down, but the percentage of prospects reacting positively to a campaign has never been higher.

Data mining enables designers to anticipate and understand client behavior better than ever before. Now they can extract the exact characteristics that determine a client's propensity to buy a product, and all from an existing customer database. This probability, called the appetence score, empowers marketers to target their campaigns more accurately. Yet these indicators alone are not enough. A large financial organization like Fortis Bank also needs numerous models to guide its marketing actions.

Moving on a larger scale
Fortis Bank required a specific data-modeling tool. "We already had reasonably well-performing models, leading to well-targeted marketing actions. But we knew we could derive more profit if we could somehow extract all of the useful data buried in our entire data mart," recalls de Brouwer. "We wanted to reach a more advanced stage: to streamline the creation of customer models and automate the exploitation of their results."

Data mining significantly multiplies marketing results while drastically reducing costs. "The marketer's job is to define strategies aimed at offering the appropriate product or service to the right client at the right time, and, finally, using the right media. This obviously has considerable budgetary consequences. Managers, however, have a different objective: cost-effectiveness. The only answer to this conundrum is to invest in profitable tools. And this is precisely why we've chosen SAS Enterprise Miner," confirms de Brouwer.

The application gives new meaning to the concept of useful data. When performing a data mining exercise, marketing analysts can type in search parameters and query the entire database for specific customer profiles. The standard SAS application immediately generates reports that contain the characteristics of customers meeting the required profile, for instance, for a specific product mailing. The application is Web-based. This means rapid and effective sharing of all the data mining results among various individuals and departments.

When programming becomes unnecessary
Experienced SAS users as well as beginners can make customized applications. The economics of this are particularly important. There is no need for manual coding with SAS Enterprise Miner. Business analysts can develop and adapt models themselves. Notes de Brouwer: "The modeling tool can be used without SAS or any other coding language knowledge."

Thanks to its SEMMA approach , Enterprise Miner dramatically reduces the complexity usually required for model construction. This facilitates the transfer of competencies inside Fortis Bank considerably. SEMMA is an acronym for Sample, Explore, Modify, Model, and Assess. It encompasses all five primary steps needed for developing models and is part of a complete set of supporting tools. It uses a structured approach that follows a process flow diagram concept. Such an organization also serves as self-documenting templates, enabling iterative data mining methodology.

Designed to satisfy everyone
The first argument in favor of SAS was its high level of performance regarding data quality control. "When speaking about model industrialization, the quality of information is the most important factor in selection. Quality is a fundamental requirement, since it affects each single aspect involved."

The SAS interface contains a data quality application that controls the entire information chain. Numerous types of tests are applied to the data, from initial entry into the database to their statistical processing. Detection of errors is automatic, and is clearly displayed in the overall shape of the tables. This allows marketers to improve the relevance of their campaigns dramatically. Moreover, this results in considerable cost reduction due, among other things, to the elimination of useless mailings, unnecessary recalls, or inappropriate contacts."

A supple framework
SAS Enterprise Miner has also distinguished itself with Fortis Bank by its incomparable flexibility. The bank covers several activities and business lines. The main difficulty in this task was taking into account the bank's diversity. As a result, the data mining solution had to answer many divergent demands, as well as being a common interface integrating all types of existing information tools. "The relationship we had with our SAS account manager greatly facilitated our mutual comprehension," points out de Brouwer.

Shaped as multi-purpose software, Enterprise Miner is a guiding but adaptable framework for conducting data mining. The user-friendly software works both in a client-server environment and in a host context with a huge volume of information. The programming phase remains the same. It enables the development of interactive applications on one platform before making them available for other situations. It also provides expanded possibilities for workgroup collaboration. Such a sharing reduces IT expenses notably.

Drill into the data
Since model industrialization demands important preparatory work – such as the definition of business, target population, model construction, and normalization – Fortis had to be sure that the solution could deliver easily exploitable and accurate statistics.

"We have yet to face any true difficulties in interpreting the figures yielded by the program. However, if that should ever happen, I'm sure we can rely on the SAS team that has been put at our disposal. They continue to offer us timely expertise," concludes de Brouwer.

Copyright © SAS Institute Inc. All Rights Reserved.