Nestor, Inc., is touting its neural network technology, so far used for various risk and fraud management applications, for in-house marketing purposes. But whether the power-hungry system can provide greater returns than ordinary methods remains to be seen.
The company says that this new desktop product, called CampaignOne, can fine-tune mass-customized credit card products by applying customized algorithmic models to portfolio-specific and even individual-specific data, allowing the firm to adapt offers to the customer while reducing costs. It claims the system can increase campaign returns by up to 20 percent.; no returns were available.
Nestor's financial services applications have so far been mainly applied to issues like fraud and bankruptcy prediction, an purpose to which the software is well suited given its ability to learn from experience, and to identify specific behavior by comparing it to typical behavior in a large population sample. Licensing CampaignOne costs about $275,000, plus maintenance costs, based on portfolio size.
"What we're trying to do is offer customers a way to do test and control groups with our models, or other models (like) logistic regression or scorecard models," says Sushmito Ghosh, Nestor's vp of financial solutions. The system, he says, can identify successful communication channels, monitor campaign performance, and continually improve campaign performance- on the fly-as the data accumulates.
Whether Nestor can really deliver this result is a matter of opinion, says Bill Bradway, director of research at Meridien Research. "It's going to be a challenge," he says. "You're dealing with a tremendous amount of data, and processing efficiency is a requirement." What Nestor is offering, he adds, is a combination of data mining and campaign management capabilities in one solution-things some large institutions can do as separate tasks, but rarely combine.