Precision marketing strikes depend on data

DALLAS - Dallas Teachers CU's Chief Information Officer Jerry Thompson sounds more like a U.S. general planning a high-tech missile attack on a rogue nation rather than a credit union executive planning a marketing strategy. Dallas Teachers recently received approval for a community charter that allows it to target some 2.2 million consumers in a five-county area in and around Dallas. To help win approval from the Texas Credit Union Department, the CU used business intelligence and spatial mapping software to show regulators just how it planned to serve the 2.2 million potential members. "We did data analysis of transportation data, our transaction data and other pieces of data to do distance/time studies. How many miles does that member live from their brick and mortar branch? Is it five minutes away, eight minutes, twelve minutes?" asked Thompson. Knowing the time/distance between members and branches is vital for Dallas Teachers because its analysis of its transaction data found that the CU has very good checking account penetration for members that live eight to ten minutes away from a branch. But anything beyond ten minutes and penetration drops off dramatically. Thompson said in order to get an accurate look at how far away (in minutes) a member is from a branch, you can't just look at a map and judge the time/distance by the physical distances. "People who live on the major arteries get to branches quicker. Some members who live 8 or 9 miles away get to a branch faster than members 3 or 4 miles away because of the roads," said Thompson. So when using a spatial mapping program, most credit unions won't be able to draw a perfectly formed circle around their branch to show where their members using that branch live. "You get very odd geometric overlays doing it this way. We had some oblong parallelograms based on where members lived in relation to their primary branch," said Thompson. The CU used this type of planning to show regulators where it planned on building branches in the next couple of years to provide adequate branch access for members in its new five-county community FOM. Thompson said the spatial software is ideal for presenting expansion plans to the CU's board of directors. The board wants to see visually where people live in relation to the CU's branches, he said. The credit union recently used spatial mapping of traffic poll information to alleviate an overflow situation at one of its branches. The older branch was having trouble handling the everyday member flow. "We wanted to know how much attrition or cross over there would be between the new and old branch. Doing the analysis on the new location we found that if we built that new branch within a 4 to 5 mile radius from the existing one we would have between 22 to 25% attrition, which is in the range we wanted." Thompson said the new branch went up in November and the attrition rate is about 24%-dead on based on the mapping software. The credit union is now using spatial mapping to find out where potential profitable members live. Looking at its own data warehouse and doing a study of what its top 10% of profitable members look like, the credit union compares that with public record data and data from third-parties to find where potential profitable members live. "We mapped these out, and what we found were clusters here and there of profitable consumers. That is they fit the profile of our top 10%." Like a military operation attacking a vulnerable point of an enemy, the credit union looks at those clusters as its main marketing targets. Now that the credit union has the locations of the clusters it can take its traffic time/distance analysis to determine the most optimal places to build a new branch to serve as many consumers in the profitable cluster as possible. Dallas Teachers uses an IBM DB2 Universal Database, IBM DB2 Intelligent Miner, IBM Visual Warehouse and Business Objects OLAP to compile its data on members and potential members. The CU's core processing system is powered by Concentrex. The credit union is even using a technology that monitors what radio stations cars driving in a particular location are tuned to and at what times. The credit union is planning on placing radio advertisements on the most popular stations at peak drive times. While it sounds Big Brother-ish, Thompson said the radio capability does not look at who the person in the car is or even gather any information about the car. Thompson said these marketing efforts may seem a bit extreme for a credit union, but competition in Dallas is so tough, it must be done to survive. "All the big banks are here. There's a bank on every corner in Dallas. If you don't do these types of things, you're in trouble," said Thompson. High-tech tools like spatial mapping software aren't of much use to a credit union if it hasn't done extensive data analysis. Data is king today and everyone is trying to find out how to get to it and what to do with it. Credit unions have all sorts of data in their core processing system, but constantly pulling transaction data from the core system can slow that system down said Matt Chapman, CEO of Concentrex. That's problematic because speed is essential for today's debit, credit and Web transactions. "You don't want to slow down the transaction engine with these overall inquiries. We've created a sort of atomic layer, where you're consistently using it to dump information from a transaction engine into a data warehouse. The difference is we do this on a very regular basis, as opposed to older systems where it's every 30 days." Concentrex uses data libraries form Sagent and data tool sets from Cognos to help credit unions understand and utilize data for more specific marketing. "They can geocode data so we know this member was here or there at time of transaction. You want to be able to tie in all the characteristics of a member for marketing campaigns. For example, query for all the members over 50 that use certain products and live in a particular area," said Chapman, In order to do really high-tech marketing credit unions need to look at public data too, said Chapman. "As credit unions go into high-tech marketing, they'll be surprised how stunningly accurate it can be. They already have the transaction data. Using an address they can get information about the neighborhoods their members live in looking at public information. The complexities of the models are extreme. We're talking about areas of genuine artificial intelligence." Looking at different aspects of a member's life brings up the privacy debate, said Chapman. Credit unions are uniquely positioned to benefit from the privacy debate-more so than banks-he said. "People are afraid of all this information out there. Credit unions should market privacy to members and say `hey we're not interested in selling your data to third-parties.' They should give members that opt-out ability if they don't want the credit union making product recommendations that may ultimately benefit them." Chapman said consumer choice needs to come first. "When I tap into Amazon.com they make recommendations based on my buying patterns. I like that. It freaks other people out, but those people can opt out." Marc Roberson, senior vice president for credit union data processor FiTech Systems, Atlanta, said finding profitable members is one of the most important tech/business issues for credit unions today. And fortunately, the tools to do it are dropping in price. "It's much cheaper. The disk storage is a fraction of what it used to be. The speed is considerably better. The tools to mine that data are so easy to use. They can be Windows-based drop and drag. It's also much more accessible to the end user to know what he or she wants. Today's tools don't require and MIS degree," said Roberson. Roberson said FiTech is giving CUs reporting tools that allow the CU to extract data from not only their member data base transactions, but all of their loan databases, data from the credit bureau, demographic data, and place it in an MCIF to use in member profiles and one-to-one marketing campaigns. "I think one of the most interesting angles coming out of all of this is the final realization that sales versus service is really going away. Poorly done selling, wasting members time, is bad service. A telemarketer calling at 5:45 when the family is sitting down for dinner is not a welcome event. But appropriately using database technology to understand precisely what your members need and when they need it is welcomed," said Scott Hansen, vice president of software for MCIF and database software provider Harland. Harland is set to unveil Channel Expert later this year which will look at MCIF data and use it to push a marketing message out to members via different CU channels. "It's a piece of technology between the MCIF system and the channels. We might send a member on the Web a gif for a home equity line of credit if they fit that profile. If the member clicks on that gif and says `yes I'm interested' we store that in the MCIF so they don't get the same HELOC message next time. That's where technology comes in. Targeted marketing that doesn't annoy members," said Hansen. But, of course, said Hansen the data is where it all starts. -pgentile@cutimes.com