Jesus Mena
Data Mining Your Website
Boston: Digital Press, 1999.
paper, 368 p., ISBN 1-555-58222-2, $US39.95
Digital Press: ttp://www.bh.com/digitalpress/

The author, Chief Executive Officer of WebMiner (http://www.webminer.com/) - a data mining consulting firm - recently penned a well-written treatise on the method and value of extracting information about transactions on Web sites. While Mena's company's Web site leaves much to the imagination, the book does not.

Mena introduces the woes of Mine2.com (http://www.mine2.com/), a portal, in order to outline the concept of mining data; he uses clear terms that are understandable, even to novice data miners. In framing the reasons for businesses to study their Web transactions, Mena points out how the effort answers several important business questions, such as "What will visitors buy?" and "How much will they spend?" If readers are looking for justification to initiate data mining, the first chapter presents several good arguments that can be used to strengthen a business case study.

Readers will appreciate the author's easy definition of data mining; that it is the tools of pattern recognition (p. 41) rather than log file analysis (p. 57) which help marketing and business departments answer questions about trends in visitors' behavior and the characteristics of loyal Web customers. The author lists several well-known data mining tools and log file analysis tools in his explanations, as well.

The remaining chapters give background to concepts such as neural networks and CART, then present the ten steps to mining data. Data mining tools have a chapter devoted to them, as do data components to be mined and helpful external information resources (i.e., marketing and other data providers). There is also a chapter about the relationship of data mining and electronic business transactions ("e-tailing"). Plenty of diagrams, charts, and pictures help make sense of these concepts. Jesus Mena also mentions vendors, providers, software, and tools by name with objective descriptions of services.

Throughout the book, but especially in the final chapter about putting all the parts together, Mena's advice inserted between instruction is practical, appropriate, and obviously based on experience. For example, on page 307 where the author discusses using third-party information providers to match mailing addresses with other types of information, he suggests: "It is recommended that you do some comparison shopping ... Ask for specific statistics on their information content and their percent of file coverage."

A minor complaint: I would have liked to see an appendix with links to online references. (WebMiner does offer free excerpts from the book, though.) In the book, there is an appendix with listings of privacy consortia, standards, and legislation - this is where the online references would have been helpful since timely information about legislation would be of high interest. There is also a glossary with selected terms.

In short, the book is highly recommended, without reservation. Pay special attention to chapters 1, 2, and 9. The remainder will be there as a reference when you are in the throes of data mining.- Beth Archibald Tang.

You can get just so far blind. You can build a store blind, and even stock it. But to tweak it, to maximize sales, to understand your shoppers, you'll need to resort to some kind of data mining, which author Jesus Mena defines as "the iterative process of extracting hidden predictive patterns or profiles from large databases, using artificial intelligence technologies as well as statistical and marketing techniques." If this sounds like some serious number crunching, you're exactly right. But it's not the same as mere analysis of traffic logs that report on server activity, as important as that is.

Data mining looks for patterns of visitor behavior. The end product is insight about the identity and preferences of your online customers. What kinds of groupings do you find? What kinds of patterns? What do they tell you? Who will buy? What will they buy? How much will they buy? What relationships exist between your visitors and your products? These are the kinds of questions that data mining seeks to answer.

Mining your data involves 10 steps (though some might be skipped by a particular analysis): (1) identify your objective, (2) select your data, (3) prepare the data, (4) evaluate the data, (5) format the solution, (6) select the tools, (7) construct the models, (8) validate the findings, (9) deliver the findings, and (10) integrate the solutions into your marketing strategies.

Now when Mena begins delving into the tools and algorithms that underlie these steps, I'm quickly over my head. The book is deep into databases, statistical analysis, and artificial intelligence tools to look for patterns and make sense of them. As Mena notes, data mining has been around for years, and was used, for example, by Wal-Mart to optimize their stores. Even though my eyes glaze over at this level, I am delighted that such a book is now available for Web marketers to study and master.

While the book doesn't provide all the techniques and tricks of the trade, it does serve as a useful introduction. In successive chapters, the author leads you through the various steps of the process, and describes the software tools at your disposal. Then he outlines in some detail each of the kinds of useful data one can extract from log files, cookie files, forms, and the proposed open profiling system. Next he catalogs each of the providers of marketing data, and then the web services and software providers that enable personalization and one-to-one marketing systems.

After discussing in detail the technologies involved, Mena devotes a chapter to e-retailing, and just how data mining can improve customer service, the percentage of sales closed, and ultimately, the bottom line. The final chapter, entitled "e-mining," summarizes how various approaches to the data can result in a variety of predictive analyses. The book includes an appendix of privacy consortiums, standards, and legislation, and a glossary defining all of the technical terms Mena refers to.