LOGISTIC REGRESSION MODEL (LRM)
Better Targeting. Better Customers. Better Performance.
Traditional live check campaigns are becoming increasingly difficult to manage effectively. Response rates have steadily declined over the last several years while lenders continue to face increasing pressure to control default rates, reduce mailing costs, and improve portfolio performance.
D2K’s Logistic Regression Modeling (LRM) solution was developed to address these challenges.
Unlike traditional targeting methods that rely on broad segmentation and generic scoring models, LRM uses your historical customer performance data to identify prospects most likely to respond—and least likely to default.
What is Logistic Regression Modeling?
Logistic Regression Modeling (LRM) is an advanced multivariate modeling technique that analyzes dozens of customer attributes simultaneously to predict:
– Response Propensity
– Credit Risk
– Customer Quality
Rather than evaluating customer characteristics one at a time, LRM considers how multiple attributes interact together to create a complete customer profile. The result is a smarter, more accurate approach to customer targeting that is customized specifically for your lending environment.
Proven Results
Our clients have experienced significant improvements in campaign performance through LRM-driven targeting:
Typical Results:
– 150% to 600% Response Rate Increase
– 300% Average Response Increase
– 10% to 60% Mail Quantity Reduction
– 25% to 60% Default Rate Reduction
By mailing fewer, higher-quality prospects, lenders can reduce costs while generating stronger portfolio performance.
LRM Actual Campaign Results
This table shows actual results from our first test campaign in November 2025 through our most-recent expanded Spring campaigns.| Client | LRM Campaign | Total LRM Cashed Checks | Total LRM Mail Quantity | LRM Response Rate | Previous Campaign Response Rate | Response Rate Increase | Previous Campaign Mail Quantity | Mail Quantity Decrease |
|---|---|---|---|---|---|---|---|---|
| Client 1 | November | 5,815 | 67,147 | 8.7% | 1.1% | 687.3% | 75,582 | 11.2% |
| Client 2 | January | 1,379 | 24,337 | 5.7% | 1.2% | 357.3% | 82,680 | 70.6% |
| Client 3 | January Re-mail | 222 | 9,453 | 2.4% | 0.8% | 193.8% | 69,712 | 86.4% |
| Client 4 | January | 1,536 | 28,069 | 5.5% | 1.6% | 243.8% | 15,052 | -86.5% |
| Client 5 | January | 844 | 30,079 | 2.8% | 0.4% | 600.0% | 26,519 | -13.4% |
| Client 6 | March | 2,577 | 54,973 | 4.7% | 1.7% | 176.5% | 51,435 | -6.9% |
| Client 7 | March | 375 | 12,543 | 3.0% | 1.3% | 130.8% | 15,051 | 16.7% |
| Client 8 | March | 367 | 10,145 | 3.7% | 3.9% | -5.4% | 13,965 | 27.4% |
| Client 9 | March | 542 | 25,174 | 2.2% | 0.3% | 532.4% | 36,231 | 30.5% |
| Client 10 | March | 422 | 11,972 | 3.5% | 0.8% | 340.0% | 9,008 | -32.9% |
| Client 11 | March | 2,327 | 59,767 | 3.9% | 0.9% | 333.3% | 105,905 | 43.6% |
| Client 12 | March | 2,138 | 73,602 | 2.9% | 0.9% | 222.2% | 199,281 | 63.1% |
Why Traditional Selection Methods Fall Short?
Many lenders still rely on broad selection criteria, generic scores, and manual rule adjustments to build live check campaigns.
Traditional methods often:
– Evaluate customer attributes individually
– Miss hidden opportunities within customer populations
– Depend on broad exclusions that eliminate profitable prospects
– Require subjective adjustments and constant management
– Fail to reflect the unique characteristics of your portfolio
The result? More mail. Lower response. Higher risk. Wasted marketing dollars.
Why D2K's LRM Approach Works?
Traditional segmentation relies heavily on individual characteristics such as credit score, income, or payment history. LRM uses multivariate analysis to evaluate all customer attributes together.
This allows our models to identify prospects who demonstrate strong response potential and acceptable risk levels—even when individual attributes might not appear favorable on their own.
By understanding how attributes work together, LRM delivers:
– Improved response rates
– Lower default rates
– More accurate targeting
– Better campaign profitability
Discovering Hidden Customer Opportunities
Broad exclusion strategies often eliminate large populations of potentially profitable customers. For example, many lenders automatically exclude lower-credit-score prospects because historical averages suggest higher risk. However, even within these populations, many customers perform exceptionally well. LRM identifies those hidden high-performing segments by analyzing the complete customer profile rather than relying on a single attribute.
Benefits include:
– Expanded market opportunity
– Increased customer reach
– Reduced missed opportunities
– Greater loan growth potential
Customized for Your Business
No two lenders are exactly alike. Your underwriting practices, branch operations, servicing procedures, customer demographics, and market conditions all influence customer performance.
That’s why D2K builds every LRM solution using your own historical response and performance data.
While our models incorporate traditional bureau scores such as FICO and other industry metrics, we go beyond generic scoring systems by determining:
– Which attributes matter most to your organization
– How those attributes interact together
– The optimal weighting of each factor
– The best balance between response and risk
The result is a model uniquely tailored to your business.
Direct Mail Campaign | Performance & Response Tracking | MLA | Data Processing