Deployment strategies for increasing your data mining ROI
PART 1
Source: www.spss.com
Copyright SPSS, Inc. 2004


Success, success and more success are the responses from organizations that have realized the benefits of data mining. They discover information they can use to increase profits by finding new sources of revenue, making processes more efficient, identifying cross-selling opportunities, targeting better prospects, reducing risk, detecting fraud, the list goes on and on. So what exactly is data mining? There are a number of definitions, but most involve using analytical techniques to discover useful relationships and patterns in large datasets.

Data mining enables you to make better decisions by leveraging both your data and your expertise. It gives you the confidence to make smarter decisions based on historical data, not gut feeling. With your data mining results, you can change what you do by learning why things happen, create new customer segments and make better forecasts. In short, customary reporting processes provide organizations with information about what has happened; data mining empowers decision makers to change what will happen. 

The result of data mining delivers one or more decision models. These models give you the power to understand relationships among the fields of your datasets and make predictions based on patterns in your current data. Models help identify problems before they occur, suggest new processes and deliver solutions you otherwise wouldn't have found. Once tested, decision models can be deployed throughout your organization to help people make smarter, more confident decisions.

The key to data mining ROI: deployment
Deployment, put simply, is distributing data mining information in a usable format to the place it can be used most effectively. Today, deployment typically comes in the form of printed reports given to decision makers, or customer lists handed off to a mail house. While distributing reports and lists are important deployment methods, strategic deployment takes you much further.

More technologically savvy organizations are distributing results on an intranet. Some have customization features that, for example, allow a regional manager to see results from only his or her region, while cutting the clutter of information from other regions. The key to success is sharing your data mining results with others in your organization - those who can use this powerful information to make positive, profitable changes and achieve your target ROI.
Deployment strategies for increasing your data mining ROI
To get the full value from your data mining investment, you must strategically deploy your models. Currently, most attention in data mining is focused on the analysis. While getting good answers is important, it's only by using results that you meet your data mining objective - whether it's reducing fraud, retaining your profitable customers, targeting better prospects or increasing productivity. 

Strategic deployment involves integrating models into your organization's operations. Model builders, as in the past, create models, but now others throughout organizations use them. Using deployed models, a front-line manager has the ability to feed new information into models to get results that boost returns in day-to-day activities.

Today, many organizations aren't getting the highest possible returns from their data mining investments. Fortunately, software is beginning to meet the requirements for successful deployment: easy delivery to a wide group of decision makers and instant updates to keep up with a fast-paced environment that constantly requires new models.

Deployment ideas that make a difference
Here are just a few ways strategic deployment improves the return on your data mining investment:

1) Customer Relationship Management (CRM). The goal of CRM is to maximize profits over time by acquiring, cross-selling, up-selling and retaining customers. This goal is met through automation and integration of front office business processes. These involve sales, marketing and customer service via multiple, interconnected delivery channels, including direct mail, telephone call centers, the Internet and physical retail locations. 

Data mining is the heart of CRM. It delivers models that open the door to discovering who your customers are, what they value and how likely they are to respond. To successfully acquire, cross-sell, up-sell and retain your customers, you must know the history of your customers, their needs and what they value. Data mining finds these answers by calculating the "lift" - or best response ratio - in a customer acquisition effort, identifying customers who would most likely be interested in other products and services, and discovering why customers remain loyal. Effective deployment delivers results throughout an enterprise so every point of customer interaction, or "touchpoint," can use models to make better, more personalized responses.

Deployment strategies for increasing your data mining ROI
2) e-CRM. The focus of e-CRM is a key customer point of contact - the Internet. Mining and deploying Web data delivers true one-to-one, real-time marketing. Two uses that offer great potential are right-selling in Web stores and serving custom content. Right-selling includes cross-selling, up-selling and other tactics to increase a customer's lifetime value. As customers place items in their baskets, models pinpoint other items of interest, making the right offers to ensure shopping carts aren't abandoned and customers remain happy. 

Models that provide custom content turn static sites into intelligent sites by analyzing click behavior. Using tested models - based on historical data - visitors to your Web store can receive special discounts, offers and relevant announcements based on their click patterns. Information overload on the Web disappears when your content is tailored to each visitor - your customers stop searching and start finding.

3) Accurate forecasting. Deploying your forecasting models stops the cycle of inaccurate estimates by replacing forecasts once based on gut feelings and guesswork with accurate data mining models, based on historical data and specific business knowledge. By deploying forecasts, business managers use the same, accurate models and adjust factors they control in order to fine-tune the forecast for better planning. 

4) Fraud detection. Data mining models can prioritize likely fraudulent cases and identify cases that yield the biggest savings. While a central office most often has enough data to find fraudulent patterns, local operations are the best places to stop fraud in its tracks. Deploying data mining results to local offices can stretch scarce resources. A sound deployment strategy is becoming even more critical with the Web, which opens new ways to commit fraud. New types of Web fraud happen every day, requiring new models to be deployed constantly. 

5) Value delivered more efficiently. Manufacturers deliver value through an extended chain of customers, dealers, suppliers, processes and products. Data mining discovers new opportunities to improve processes. When you deploy data mining results, you build a nimble enterprise that predicts breakdowns before they happen and resolves inefficiencies in the supply chain.

Deployment strategies for increasing your data mining ROI
6) Risk analysis. Health care, education, finance and other industries need to assess risk: the effects of treating a patient, offering a student a place in class or extending credit to an applicant. Data mining models effectively identify risk. The decentralized decision making in risk analysis makes it an ideal application for deployment. With deployment, any organization - from global enterprises to small institutions - can consistently manage risk.