Larry Filler knows your fear. He’s ready to talk you down.
As a partner with Boire Filler Group, a database marketing consulting agency, Filler helps organizations of every kind try to sift through reams of customer-related
information and turn it into a blueprint for more successful marketing strategies. In other words, he knows about data mining, a concept that both intrigues and unsettles senior marketing professionals.
Filler’s expertise was developed in a variety of roles. He spent three years in the advertising industry at MacLaren McCann Relationship Marketing, where he worked with clients like General Motors and the National Post. At the Loyalty Management Group, he advised clients on how to use the Air Miles Reward program and its database to their competitive advantage. Filler also spent six years at American Express developing marketing strategies for the company’s insurance, credit card and merchant businesses.
Filler recently helped Pipeline answer some of the most common questions about how to go deep into the data mine and how to come back with something useful.
Pipeline: Many companies have a lot more data than they need. How should marketers begin a data mining project?
Larry Filler: With a lot of our clients who haven’t done anything, we call it a “”data audit.”” Sometimes you have to massage the wording that you’re using with a client, but what you’re doing is exploring, just looking at some of their data. You’re also, before you do that, trying to understand what their business is and what they think they want to use the data for. Sometimes they may not know that. For example, if you go into someone’s data and things like age instead of date of birth, well, sometimes that’s not as valuable, because when did you collect the person’s age? What if you have 100,000 customers and you only have age for a third of the people? Then that variable isn’t going to be as useful, because 65 per cent of your database doesn’t have age. That might have them look at how they’re collecting that data. There may be ways to append external information to get people’s age and things like that. It’s going through that process.
Pipeline: How do you really turn that kind of data into something that will help you improve as a marketer?
LF: Some people really don’t know who their best customers are. They haven’t even started to stratify their customers just based on value. They know that they’ve sold X number of product last month and they spend X number of dollars on advertising and they got X amount of revenue. That’s great, but is that new customers coming in? Is that the same customers buying more? You need to shift away their thinking slightly in terms of analyzing it because the reality might be that they’re spending more on advertising, but that advertising is not getting more individuals to buy. Or, conversely, they’re getting more people to try it, but then they never use it again, so that month the sales looked really good, but it’s not sustainable.
Pipeline: A lot of firms are using advanced customer relationship management tools and techniques to tackle those challenges. Is CRM helping?
LF: I think CRM, or the work that we do in terms of the analyzing data, began to be driven by the technology. People were buying the tools to do this kind of work but didn’t realize you needed the strategy around it, or you needed people who were going to use those tools. A lot of organizations have made large investments in technology and haven’t seen the fruits of that investment, because they still need budget to use it. I think there are more pragmatic ways you can go about it before you have a big investment in some of these tools. There are ways to do things that maybe aren’t as automated or aren’t necessarily enterprise-wide at first. I do think a lot of companies are thinking differently today, but I think their challenges are still implementing some of their solutions.
Pipeline: Is the push towards data mining coming from the marketers, or is that something that’s being foisted upon them by senior management?
LF: It’s probably coming from the marketers. People realize the value of the data. Ten years ago, if you said “”data mining,”” no one knew what it was. Today that’s no longer an issue. They understand the concept of what it’s supposed to do. But I’m still not convinced that all the organizations are doing it. But the hurdle of “”What’s that?”” isn’t there anymore. Now it’s, “”Is it a priority for them?”” and as you said, “”How do they get started?”” And then how do they integrate it into what they’re currently doing? If they’re doing direct marketing, it’s easier for them to understand the savings opportunities and the better ROI. If they’re shifting from like a retail organization that has mainly done general advertising or some flyers, it’s a little more difficult.
Pipeline: If people don’t need to necessarily buy expensive tools at first, what should marketers expect in terms of the time and human resources allocated towards data mining?
LF: I think that’s going to be more determined by the scope and scale of which they’re going to do things. It’s not unreasonable to spend between $15,000 and $50,000 to try and get an assessment of where you can go and get some initial tools developed. That wouldn’t be out of the ballpark. We would go to a client and some of the work that we might do is bascially review their data, the quality of it, look for any trends, as well as do some value segmentation of their customer base. Even doing things like some quick RFM analysis on their data that they’ve maybe never done, to tell them if they have an 80/20 rule happening. Some people don’t know that. Just understanding that can be eye-opening to a client. Or retention rates: How many people bought last year didn’t buy this year? Those types of things, I think, can be very helpful in getting them started and getting them to understand the value. Sometimes just getting things done internally in organizations is difficult. If it’s not a priority, and operational issues in their organization become a priority, then that piece of the project gets put off. Sometimes using external resources ensures that it gets done.
Pipeline: Would it be fair to say organizations ignored data mining when sales were good but that they’re more interested now because we’re trying to get through a slump?
LF: I think there’s some truth to that. If everything’s driven by an ROI . . . let’s say when the economy’s good you were going to get a three per cent response rate and when the economy drops you’re going to get a two per cent response rate. There’s still a segment of people that may still be responding at three per cent, so you want to stratify your customers. If you looked at the lifecycle of a product, the data mining can become more meaningful and the segmentation can become more meaningful as your sales start to shrink. So, for example, if we build a tool for somebody, maybe initially they’re still going to 50 per cent of the list. But then a year later they’re going to 30 per cent of the list because response rates are dropping if you’re offering the same product to the same people. So you get more focused. But there’s also opportunities for customers to start using this for product development. Like, here’s a group of people that don’t buy these products. Well, what sorts of products and services should I develop for them?
Pipeline: So once they’ve done data mining, are marketers more savvy about how to change their strategies?
LF: I think that’s where it’s much more long-term. People need to be committed to it. But the key is, people need to be able to understand their internal challenges. If someone has done no data mining or no analytics, there are probably some quick wins. You need to demonstrate those quick wins and show here’s why you should invest more money in this area.