Predictive modeling is one of the most effective tools that organizations can use to target a qualified audience. In essence, it is the practice of using predictive analytics – data mining that indicates the potential for upcoming trends – to form a statistical model of future attitudes and behaviors. Using this information, organizations can form ad campaigns around the anticipated needs of their consumers.
In one case study, a Midwestern electric company decided to use predictive modeling to help lower-income families get access to the financial assistance they need to pay costly winter heating bills. In the cold months of November through March, it’s illegal for a company to turn off a resident’s power, even if the bills can’t be paid. For many heating and electric companies, this means losing out on millions of dollars of profits for services rendered. It might seem like an impossible solution, but the electric company in this case study decided to get creative and participate in a little predictive modeling.
The Low Income Home Energy Assistance Program is a federally funded program that provides cash grants to help people pay their energy bills in the winter months, when they’re likely to skyrocket. The electricity company engaged in predictive modeling to target 23,829 potential candidates who had no previous indication of being an energy assistance participant but who, according to predictive data, could benefit from the program and would be likely to participate.
Using the data gleaned from the predictive model, the company mailed out information packages and applications for the assistance program. They received an incredible 6.95 percent response rate. Based on an average grant of $1,000 per household, this equated to a potential return of $1,657,000 – which was nearly a $59 return on investment for every dollar the power company had spent in their direct mail campaign.
The end result was good for both the company, which was compensated for services rendered, and the targeted audience, who were able to obtain the assistance they needed. It was also good for the federal government, which had a better idea of the actual need for the program within the district and can predict need with greater accuracy in the future.
This is just one example of how predictive modeling can benefit a campaign for the good of all parties. It can also help non-profits determine the best candidates for giving over the holiday season, when budgets are tightly strapped, by targeting affluence data. It can be used to help businesses reach just the right audience for a new product launch and may even indicate new sectors that have a need for older products and services. Whatever the situation, predictive modeling can be used to target the individuals most likely to respond.