How To Mine All That Customer Data

The seeming simplicity of a term like “business intelligence” belies the rich potential its implementation can provide to a small or midsize company. Data mining is the backbone of BI–it means sorting through company data and identifying and extracting valuable information about operations. In a customer-centric business, the information can be particularly compelling. We turned to Gordon Linoff, a principal at Data Miners, to learn what data mining for BI can do for small and midsize businesses.


Linoff: Once upon a time–and this is a rosy view of the ancient world–the hardware store owner knew you. If you went to the grocery store, someone there knew what items you needed. If a woman walked into a bank to make a deposit and she was pregnant, [the bank manager] knew and offered her a car loan. We traded all that for lower interest rates and lower prices, we traded all that for more efficiency, but we lost the intimacy with our customers. How can we get to know our customers again? That’s one component of business intelligence.

Where do small and midsize companies fit into this BI picture?

Well, it depends upon their relationship with their customers. If they only have a handful of larger clients, they know them. But one of our clients, The Vermont Country Store, is a cataloger that sends to several million people. The company has about 400 employees. How are 400 people going to know about several million people? They need to use the information they collect in order to better serve their customers and increase revenue.

But there’s not an overall approach, because businesses are different, and there are several different levels that businesses operate on. Some are campaign-driven, to encourage customers to buy, and [business intelligence helps] them optimize each campaign and [some use] traditional predictive models associated with data mining, tailoring aspects of their Web sites to their customers or sending e-mails to their customers to return to their site.

For example, on, which is not a small company, if I bought Harry Potter IV, [they know] I will buy Harry Potter V, VI, and VII, [because they’re asking] what’s the next type of thing that I’ll buy? You can get associations or market basket analysis. For companies in retail, association rules are something they’re interested in. What things sell together?

How can a company’s IT managers maximize the benefits of business intelligence?

The IT manager needs to enable communication–it’s the most important thing. In most companies, when [management] asks questions, [the IT manager] says it can’t be done, because people get trained not to make decisions off of data. For instance, if the marketing department of a company wants to plan a promotion, and needs to know what other items sold last year when soup was on sale, the IT manager might say, “we’ll know in three months.” So the question stops being asked and opportunities are lost. Then people start making decisions based on intuition and they [lose] the ability to make more informed decisions.

How can they get that ability to make more informed decisions?

It’s the people on the front line–the worst-paid people–who know the customers. You need a mechanism to get that information out. I have another great example. There was a guy who owned a chain of drugstores that sold chocolates on Valentine’s Day. It was big business for them. Using their BI system, the SAS OLAP server, they sliced and diced their data to discover that two of the stores had higher sales in chocolate. Both were managed by the same person.

They discovered that instead of placing the lower-priced chocolates on a lower shelf and the more expensive chocolates higher up he was mixing them together. It turns out, guys were coming in and reaching for the cheaper stuff and then saying, “she’s worth it.” This demonstrates that people on the front line know what’s going on, so you need to have systems to detect what they know. You need systems in place to communicate the best practices.

What else should small businesses be mindful of when trying to extract data?

Small and medium-sized businesses need data, but they also need an attitude toward data that they want to keep and maintain clean data. That’s their foundation, and it needs to be viewed like that. They need to want to maintain data about their clients. Vermont Country Store invested in data mining software because they hired somebody who had experience in the financial services industry, and she used the software to demonstrate that she could find more responsive customers for their mailings.

A pager company, SkyTel, developed a data mining group and had somebody who could program in SAS, [statistical software]. Their first project was to tackle one of the major causes of dissatisfaction among many of their clients, which was when they would exceed their minutes on their billing plan, they wouldn’t find out about it until the next month, when they got their bill. They decided to try to contact those people before the bill went out and [offer to put them] on the right payment plan.

It was called “right planning,” and it was controversial, but they set up a system where they could see how many pages customers were allowed, and how many they got, and then contacted them. It prevented customers from canceling. This wasn’t about advanced analytics, but about a change in attitude of a company. They talked to people, got information, and did what needed to be done.

What are some of the solutions out there that can help companies do something like that?

SkyTel used SAS. Most small and medium-sized businesses probably can’t afford a customized solution. If they’re making the investment, they need a powerful server they can use for reporting and for analytics purposes, and then they need the software on it. They can use a relational database like Microsoft SQL server or Oracle SQL, or [the open source database] My SQL.

Once they own their data, there are things they can do with it, if somebody also knows how to use Excel. But there are vertical market solutions. The statistical software [to choose from] is Minitab or SPSS, or SAS. If a person has more than 100 or 200 employees, it’s not unreasonable to have a statistician or analyst on staff.

How can management work with IT?

The IT guys need more resources. They need a copy of their operational data that’s used just for decision support and business intelligence. In any business that deals with lots of customers, computers are recording the transactions. You can’t go back to these systems, because they’re running the business, and you don’t want to interfere. So you need another copy that can go back and you can ask it questions. If you are running a business off the Web, someone else is running your Web site, so you need a copy of the systems so the IT guys can get to it and do something with that data. The business guys need to ensure that that’s funded. You need a dedicated system.

How can the return on investment (ROI) in business intelligence be determined?

To me, it’s the way a business is run. For example, Liz Claiborne had fresh flowers in her office every day. What’s the ROI on that? It’s a commitment to how a business is run with respect to customers. But there are certain applications where you can know the ROI. It depends upon the environment. If you’re sending a newsletter once a month, and it’s identical for everyone–and then you go and customize it based on segmenting customers into different groups and it drives sales, there’s the ROI.


Sourced from


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

Blog at

Up ↑

%d bloggers like this: