When you want to increase campaign response rates and generate more revenues per customer, solutions may be as close as your own customer data file. The challenge is figuring out what data predict which buyers will meet your business objectives.
That’s why more companies are turning to predictive analytics. By developing predictive models, marketers can realize that lift in response by at least 20% over more traditional list-building methods, such as income-based selection or geographic clustering.
Predictive modeling basically involves analyzing data from past marketing campaigns or other transactions to determine how various combinations of attributes affect response or purchasing behavior.
Each prospect or customer is scored based on demographic and lifestyle characteristics, as well as buying history. The marketer then targets the top scorers and eliminates unlikely responders and low-value customers.
To determine if modeling is practical for your company or for a specific campaign, you need to consider these questions:
What do you want your model to accomplish? Predictive models can help:
• Improve response rates to specific promotions;
• Determine which consumers are likely to be repeat buyers;
• Identify customers for upselling or cross-selling;
• Reduce cost by eliminating unqualified prospects or non-responders; and
• Rank leads for follow-up via telemarketing or other methods.
Do you have enough good quality data to develop a useful model? To create a response model, you need a list of about 2,000 responders and 2,000 non-responders from a similar campaign. You will also need demographic and psychographic attributes for both responders and non-responders. If you don’t collect that data yourself, you can purchase it from other data providers.
Will you have multiple opportunities to use your model? Predictive modeling is an investment. To maximize your return, you should plan to refine, refresh, and reuse a model multiple times.
Do you have the expertise to produce relevant models? Although software vendors sell sophisticated analytical tools, it often takes significant expertise to use these programs effectively. If your company lacks the resources to build in-house analytic capabilities, consider outsourcing to a consultant or data company with experience in your industry.
When predictive modeling is too expensive or impractical, consider using “shelf models.” We’ve seen financial companies use well-designed shelf models to reduce costs per lead and cost per sale dramatically, resulting in an average jump in profits of 65%. That’s a result any marketer would be happy to take to the bank.