Data Mining: Getting The Most From Your Website Data

Data mining has gotten too complicated to be useful for the big boys, but we can learn from their mistakes

Among the many ways that life in the information age has changed business thinking has been the evolution of an entirely new (well, not really) field of business management called analytics. As I mentioned, it’s not really new. Businesses have been collecting data, from Nielson ratings to telephone surveys and those lovely people you meet in the mall with their clipboards and their questions, for a long time.

The change wrought by the Internet has been in the scope of business ability to gather that data, the accessibility of it from various sources, and the capability of gathering data about any given individual from a variety of disparate sources and put them together into aggregated information. This is the ability that gives consumers the willies, because of its implications with respect to the privacy of the individual citizen. Granted, for example, court records are public. But the ability of a company to put together an entire profile of an individual, from their home address to their shoe size, the kind of underwear they bought for their wife and the driving ticket they got back in 1985 is enough to send most people into strong convulsions.

Businesses don’t do that, of course. For the most part, they have no reason to do that. And, according to a recently released report from Forrester Research, many of the businesses that have been looking to the Internet to collect data on their customers have their hands full as it is.

It’s rather comical to learn now that those big online businesses need to simplify what they are trying to do with their data if they want to get anything useful out of it. Those of us with smaller businesses have been keeping it simple, although we do so out of necessity. Our budgets do not allow for complex data collection and analytics, and our websites are often not sophisticated enough to allow us to collect quite so much data to begin with.

In addition to what we find out about our customers when they place orders with us, the kind of tracking services most of us use can tell us things like which product pages are most popular, which search terms are used most often on our sites, and where our visitors are coming from. This stuff is quite valuable for our purposes. It tells us which products we should push, what kinds of products we should stock and which of our marketing efforts are proving most successful.

As usual, however, when Forrester gives advice to Global 2500 companies, we should all sit up and listen if it gives us insights into things we can do with our much smaller businesses to improve our bottom lines. In this case, there are some principles of data analysis that we would do well to make note of.

For starters, there are two things you need to be able to do when you collect data on your customers in order to make that data useful for you. You need the ability to drill down and you need the ability to manipulate. Drilling down means that you can take your data and zero in on specific nuggets of information in order to ferret out particular individual trends and/or customer characteristics. And, of course, you need to be able to manipulate those pieces of data and put them together in different ways in order to answer different specific analytic needs as they arise.

But here’s the real gem in this particular report. Analyzing data should be a step by step business process and Forrester has been kind enough to outline those steps for us.

Model the business problem/question

Obviously, the first thing you need to do when considering an exploration of the data you have on your customers and their use of your site is to figure out what it is that you want to know. Sure, it can be entertaining and even occasionally insightful to meander through your statistics, but in the interests of efficiently using your time, you need to start out by figuring out what you’re looking for. Makes sense, right?

Prepare the data

Forrester’s vision of this step in the process is much more complicated than mine, but that’s because the odds are that our business questions will be smaller and require less complicated configurations of data than what they have in mind. That said, this is the point at which you collect various pieces of data from the various sources you have available to you. If your question involves the relationship between site traffic, inventory and sales, for example – maybe you want to figure out how to convert more lookers into buyers. You might want to take a look at specific popular items in your inventory, how many people are looking, what percentage of those people are buying, and how they are finding that particular product (website search? product category links? bookmarks? email marketing click-throughs? search engines?). You’ll also want to look more closely at the people who are looking but not buying, to see if you can figure out what is stopping them. Do they continue on to your shopping cart, only to abandon it at some point in the process? Do they continue to browse through your other products and then leave without buying anything? Do they look for more information on the product, but fail to find it? You can get all this information, but you’ll have to collect it from a number of different sources.

Analyze the data

Once you put all that together, it’s possible that you’ll discover all kinds of things about your customers’ browsing and buying habits that you never realized before. Research is often like that; you may start out with a hypothesis but it’s always possible that the data will demonstrate that you were way off the mark. You might find out that, while you are following the popular wisdom about how to merchandise your products on your site, popular wisdom doesn’t seem to be working the way it’s supposed to. Your data will show you that you need to do something very different and, more often than not, your data will also show you just what that something could be.

Forrester includes a fourth step in the process that doesn’t apply to most small businesses, which is the step of distributing the information to those people in the organization who need to use it.

This kind of thing can be time consuming but, unless you are going to blindly do as you’re told by the Internet marketing guru du jour, you need to include this kind of market analysis in your time budget. This process is what Forrester calls turning information into insight, and gaining insights into what people buy and why can be invaluable. There will always be people who are willing to tell you how you should market on the Web but the fact is that smart business means letting your customers tell you instead. All things considered, they are the folks who can tell you the most about what is working best and, in the context of your business and your product, they are really the only people who matter.

© 2000 WAHMPRENEUR News Magazine