5 Ways to Cut Costs with Predictive Analytics

SAN FRANCISCO — Have you ever heard the phrase "Let sleeping dogs lie" in casual conversation? Does "If it ain't broke, don't fix it" ring any bells? Well, those all-too-familiar cliches can often be applied to customers. Certain customers, for example, are utterly content in the present. If a customer matching that description is sent what's known as a "trigger offer," it can disrupt their prior contentment -- which might inadvertently promote churn. But how do you know which customers not to trigger? How do you segment your customer base and cut costs at the same time? This scenario was brought to light here at the Hotel Nikko in San Francisco this morning, as part of the first-ever Predictive Analytics World conference, when Dr. Eric Siegel, president of Prediction Impact and the event's program chair, delivered an opening keynote that outlined some best practices for predictive analytics, and identified the value each can provide.

"Your organization learns from its collective experience," Siegel said, explaining how models for predictive analytics rest atop existing customer data, gaining insight from information your company already has. Because of its base in operational data, Siegel told the crowd, predictive analytics -- a business intelligence technology that produces a predictive score for each customer or prospect -- has the most potential to positively influence operational decisions. And despite acknowledging a hatred of bullet-point lists, Siegel shared with attendees a list of nine applications for predictive analytics -- in precisely that format:

• response modeling
• customer retention with churn modeling
• product recommendations
• online-marketing optimization
• behavior-based advertising
• email targeting
• insurance pricing
• credit scoring
• lead scoring

And the list keeps growing, Siegel said -- as does the list of interested verticals, which has expanded from the set of early adopters (financial services, e-commerce, healthcare) to include consumer services, high technology, insurance, nonprofits, publishing, retail, and telecommunications. Although each industry has unique needs, Siegel said, predictive modeling crosses industry lines -- and brings potential cost-saving benefits along with it.

Siegel's top-five list for lowering costs with predictive analytics:

1. Don't contact those who won't respond.
Response modeling -- This type of modeling basically means targeting fewer people without sacrificing a high response rate. By modeling on top of historical customer data, you can see who is likely to respond to an offer, and then target accordingly.

2. Don't contact those who would have made a purchase anyway.
Response-uplift modeling -- The next step after response modeling involves segmenting out those who would likely purchase regardless of an offer. Having identified this segment, you can save money by removing them from the campaign.

3. Don't waste expensive retention offers on those who will stay anyway.
Churn modeling -- "Retaining customers can be quite expensive," Siegel said. "You can't afford to offer [retention incentives] to all of your customers. If you predict those most likely to leave, you can target much more efficiently."

4. Don't offend those who would otherwise stay.
Churn-uplift modeling -- That cliche is true: It's sometimes best to let sleeping dogs lie. Some customers would never have taken their money elsewhere until you reminded them you had it. It's better just to leave these stagnant customers alone. You don't want to remind them that they're paying you, or annoy them with unnecessary offers. In order to do this, you have to model a control set of customers to properly segment -- and to measure the value of modeling initiatives.

5. Don't acquire "loss customers."
Risk modeling -- You wouldn't lend your credit card to a convicted thief, would you? By the same token, you don't want to sell to customers who lack the means or the intention to pay. Risk modeling is especially crucial for credit-card providers and insurance companies, where if things go wrong the provider eats the cost.

Siegel added that there are still other cost-saving measures that can be achieved through predictive analytics: fraud detection, application processing, and supply-chain optimization, to name just a few.

News relevant to the customer relationship management industry is posted several times a day on destinationCRM.com, in addition to the news section Insight that appears every month in the pages of CRM magazine. You may leave a public comment regarding this article by clicking on "Comments" at the top; to contact the editors, please email editor@destinationCRM.com.

 

Related Articles

The Next-Generation CMO Knows Something You Don't Know

destinationCRM.com Exclusive: A new report from Forrester Research explores the maturity of your customer intelligence — and reveals what your marketing department isn't measuring.

Online Retailers Actually Had a Holly, Jolly Christmas

eTail West '09: With revenue down, retailers are having to rely more on analytics than ever before.

Welcome to the Customer Data Revolution

Predictive Analytics World '09: A former Amazon.com executive shares the steps required to manage today's customers.

Analytics the Customer Can't See But Will Experience

SAS Institute broadens its customer intelligence with a new analytics solution.

The Power of Foresight: Top Companies Use Predictive Analytics

Better customer information leads to higher retention rates.

Building on BI with Predictive Analytics

Information Builders Summit '08: Information Builders strengthens its WebFOCUS platform, delivering on cost-effectiveness and demands for analytics.

Analytics Proponents Are Often All Alone

SPSS Directions '07: Panelists advocating statistical solutions at the company's North American conference seem to be mavericks of the business industry.

Predictive Analytics Foresees Change in the Future

The keynote presenter at SPSS's Directions 2007 North American Conference speaks about gaining a competitive edge by doing the math.

Predicting Profitability

After years of trial-and-error, enterprises are finally developing innovative strategies and incorporating new software to allow them to identify -- and sell to -- their most profitable customers.