Review: Data Mining Your Website

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.

I doubt that solo entrepreneurs will have the resources to use this sort of book, but I see it as essential reading for Web teams at both small and large companies that are developing the analytical resources needed to compete in the rapidly changing online market.