Most companies have terabytes of computer data archived in a variety of databases scattered throughout the enterprise. Hidden among all that data, like three ounces of diamonds in a hundred tons of coal, are answers to questions like: 'What caused the workflow bottleneck on line A?' or 'Why are sales down?' Finding these answers can be more valuable, and far easier, than mining coal to find diamonds.

Definitions of data mining are often entangled with data warehousing. Data warehouses and data mining are complementary. Data warehouses are for storing data; data mining turns stored data into knowledge.

Business leaders are discovering that Enterprise Resource Planning (ERP) systems can improve operating efficiency, but don't provide strategic information necessary to grow the enterprise. Data warehouses store the data, but lack tools to 'mine' the jewels it contains. Reporting and Online Analytical Processing (OLAP) tools find answers to 'what' questions, but getting answers to the more difficult 'why' questions requires discovering patterns among data. That's what data mining tools are designed to do. For example, instead of a user asking for a report or graph of the average variability of product X-hoping to detect a pattern-the user instead ask for patterns relating to causes of variability related to product X. The result is more informed and more timely decision making.

Just as different techniques yield gold, diamonds, copper, and bauxite, a good data mining solution employs multiple techniques to discover patterns and transform data into usable business knowledge. Discovering patterns among data (also known as Knowledge Discovery in Databases) requires hypothesis, exploration, and pattern recognition tools tightly integrated to facilitate automatic pattern discovery and to validate findings.

Data mining tools have been available for several years, but recent advancements in computer technologies permit users to apply criteria beyond the algorithms to select data mining solutions. However, before selecting data mining software it is important to assess and align company informational needs with organizational goals, strategies, and processes. Unless data mining information is delivered in a context consistent with business direction and needs, it will be difficult to efficiently and effectively use new information. Once informational needs are understood, considerations in selecting data mining software tools include:

• Ensuring nonprogramming users can easily and intuitively address data collection, quality, preparation, and transformation activities to produce clean, warehoused data in a format suitable for pattern discovery;
• Ability to provide multiple knowledge discovery and verification techniques such as genetic and statistical algorithms, rule induction, fuzzy logic, data visualization, clustering, factor analysis, artificial neural networks, and decision trees;
• Scalability, connectivity, and sharing of popular business and manufacturing applications using recognized standards such as OLE (object linking and embedding), OPC (OLE for process control), SQL (structured query language) ActiveX, Java, etc; and
• Graphical interfaces, online help, wizards, and other familiar and intuitive user technologies.

Until recently data mining was a back-office effort conducted by programmers and statisticians focused on improving product quality. While quality improvement remains an important piece of data mining activities many other benefits exist.

The saying, 'You only get one chance to make a good first impression' is especially true of e-relationships. As the speed of business accelerates more and more companies are launching e-stores, e-business, e-commerce, e-(you name it) to gain competitive advantage, meet existing customer demands, and capture new customers, but Gartner Group (Stamford, Conn.) estimates 75% of e-business projects fail because of poor planning.

Getting it right the first time includes realizing the power of data mining to predict the future by:

• Knowing your customers, their patterns, online behavior, and demographic profiles to predict and suggest specific products, services, and other areas of interest;
• Identifying how first-time and registered visitors arrive at the site, tracking time spent at each touch point, the sequence of navigation, and encouraging site personalization; and
• Identifing customers most likely to leave and why.

Data mining can turn customer and supplier activities into actionable knowledge, but the chances of success improve when business objectives and information needs are aligned before selecting and deploying data mining software.