In any industry, maintaining or improving profitable portfolio growth is key to being successful. In a marketing environment this can be achieved by:
• Increasing responses of high quality/revenue generating customers
• Maintaining responses whilst reducing marketing spend
• Excluding customers likely to generate high delinquency rates

This article provides insight into risk and response profiling and the application there of in soliciting new customers or cross-selling products to an existing customer base.

Planning and Data Quality
Campaign results are often inhibited through poor planning, product selection and targeting the incorrect audience. Specific consideration of data inaccuracies and risk selection are also often overlooked. These factors are important to any marketing initiative and sufficient preparation is required prior to applying any type of predictive technology.

Direct mailing initiatives can have a huge impact on system and operational areas and effective resource allocation needs to be considered. It is best practice to roll out pilot campaigns to test the impact on operational areas and identify system inefficiencies. Prior to a full mailing campaign, mailing quantities should be staggered to prevent day-to-day operations being over burdened. Campaigns should be monitored closely to test the impact thereof, and if the desired results are not achieved, the mailing can be refined.

In developing any type of profile it is imperative to take cognisance of the fact that the strength of the profile is directly related to the validity and amount of information available. As the majority of the project time is spent on the initial data validation and analysis, sufficient time should be allocated for this in the project plan.

Customers can be targeted from various data sources:
• Internal Customer Database
• Purchase of External Marketing Lists
• External Database

When little or no behavioural data is present, external information is available via the credit bureaux to provide additional lift in the profiles. The diagram below illustrates the general quality of information available from these sources

Certain data sources will require more preparation than others e.g. validation and cleaning/scrubbing. If one looks at internal company data one would expect good quality data and sufficient history to start mining the database. More often than not data irregularities exist and significant time is spent on getting the data right.

Profiling
Profiles are developed to predict a specific outcome throughout the credit life cycle, from the initial application through to final bad debt. For example, risk profiles predict the future performance of an existing customer, over a specific period of time. Performance measures such as payment behaviour and delinquency migration are used in this type of profile development.

A response profile on the other hand predicts the likelihood of a potential client responding to an offer. When response and risk profiles are used in combination with each other, they provide the ability to directly target customers with accurate precision. Not only is one able to solicit customers who are likely to respond, but one also knows their likelihood of going bad in the future.

Profile development times differ depending on the quality and depth of information available from the various data sources. A minimum of twelve months performance information is required to build robust risk models, although response models require between three to six months data. Generally, the more data available for the development process, the stronger the profile is.

Internal customer database
Generally, robust profiles can be developed as application and behavioural information is in abundance and sufficient performance history is stored. When combined with past campaign results, this can produce strong risk and response models.

Marketing Lists
Marketing lists typically lack behavioural information, but risk profiles can be developed using demographic information, combined with credit bureau based information (bureau scores, payment profiles, etc).

External Database
As external databases contain some performance information, the development of profiles is similar to that when using an internal database. Credit bureau information is often used to provide additional risk assessment (bureau scores, payment profiles, etc) where behavioural information is scarce.

Profile Application
After developing the respective profiles the appropriate actions can be deployed based on the risk and response profiles. A matrix approach is often used and strategic selections can now be made depending on the level of risk and the desired response rate that the company is prepared to accept. This approach enables the company to tailor its product offering and plan the impact of a future campaign on its operational areas.

From the matrix below descriptive labels have been assigned to distinguish between risk and response. The best group, which contains the low risk and high response segments, are classified as ‘Diamond’ and ‘Ore’ is assigned to the high risk and low response segments.

Selection Strategy
From the above matrix the best groups can now be targeted. Offers can be tailored to stimulate reaction from certain profiles, e.g. better offers to the low risk-low response group may generate improved response and standard offers can be reserved for the higher risk-higher response groups.

Different types of offers can be introduced from the standard ‘invitation to apply’ offer to pre-approved products. Pre-approved offers generally stimulate substantially better response rates. When used in a credit offering, the pre-approved limit offered can be tilted based on response and risk levels. E.g. Diamond groups could receive higher pre-approved limits whereas Bronze would receive lower limits. In higher risk segments, the use of invitation offers allows additional risk assessment information to be introduced as applicants can be re-assessed by the normal application process on application.

When sufficient information is known about the customer the level of service can be greatly improved by automatically approving the offer on your host system. This is useful in planning resource allocation, e.g. reduced underwriting, and improved overall customer service.

Summary
The benefits of using profiling in marketing campaigns can range from cost saving in mailing costs, to better response and conversions rates. Various strategies can be implemented using the profiles, from targeting product offers to more effective resource planning. Ideally a direct marketing plan should provide sufficient time for complete data validation, response and risk analysis, monitoring and the tracking of mailing strategies.

About the Contributor
PIC Solutions is the leading specialist credit risk solutions company in the EMEA region. With offices in Cape Town, Dubai, Johannesburg, Manama and Nairobi, we deliver integrated analytics, consulting and software solutions to over 150 companies in 30+ countries. We work worldwide with organizations to improve performance, drive strategies and enhance profitability. Analytics | Consulting | Software