David Mattei and Peter Kulik are ATM experts who work for Fifth Third Processing Solutions. Mattei, a 20-year industry veteran, is Fifth Third's product manager for reporting. Mattei focuses on a suite of products that provide access to transactional and ATM data. Kulik, Fifth Third's new technologies product manager, is a regular trade-press contributor and industry speaker. He too has 20 years of experience in the ATM and banking industries. To make a comment about this commentary, please e-mail the editor.

Peter Kulik

A financial institution recently came to us with a problem. It had moved an ATM from one site to what it thought was a better one just a few blocks away; but transaction volumes dropped and a surprising number of customers complained. As the institution's EFT processing partner, the financial institution wanted to know if we could help them determine why the problem occurred and how we could help them fix the situation for their customers.

An independent deployer had heard about co-branding alliances and was interested in exploring the strategy, to increase its transaction volumes. But there were five local financial institutions - all of which had ATMs deployed - and a regional institution rumored to be planning a branch in town.

Another financial institution was hit hard by Hurricane Katrina in 2005, and a large part of its customer base was displaced. Customers desperately needed access to cash, and the financial institution needed to keep its customer base intact while maintaining the high level of customer service for which it was well known in the market.

What do those three seemingly different problems have in common? They can all be solved using data mining.

By analyzing ATM-transaction data we can answer important business questions and provide data-based input to guide business strategy. Fifth Third has used data-mining to answer those questions, as well as assist in a wide variety of other possible business applications.

Where are the most popular foreign ATMs that my cardholders are using? That answer depends on the number of transactions performed or the amount of surcharge fees that cardholders pay. Should I, as the financial institution, deploy an ATM near some of these locations?

The financial institution mentioned in the first scenario needed to answer those questions in order to determine why transaction volume dropped so substantially when it moved its ATM.

David Mattei

By analyzing transaction data, the institution determined that its cardholders, rather than going the short distance to the new site, began using a foreign ATM located adjacent to the old site. Cardholders were paying surcharges at this new ATM, and they weren't happy about it.

By using a data-mining report, the institution was able to evaluate the ATM's location. It decided to move the ATM back to its original location to support customer habits - and maintain customer loyalty.

Another financial institution analyzed a similar report to identify the most popular foreign ATMs that cardholders visit. To the institution's surprise, one top foreign ATM was actually around the corner from one of the institution's branches. This financial institution mined the data to obtain a list of the specific cardholders using the foreign ATM and then mailed out an informational/marketing letter to those cardholders, informing them of the branch located right around the corner.

The institution was able to reduce foreign ATM interchange expense and offer surcharge-free service to customers with a very low investment in time and cost.

It's an old business adage: Know your customer. However, many ATM deployers have no idea who is using their ATMs.

To evaluate potential alliances, the ATM owner mentioned in the introduction mined transaction activity to generate a BIN-level analysis. The analysis identified which local financial institutions had the most cardholders coming to the deployer's ATMs. With this information, the ATM owner was able to approach the institution with whose customers he did the most business to explore alliance strategies.

The institution was not the largest in town - and had no plans to add ATMs to its base - but it did open several alliance opportunities.

A financial institution could use a similar report to create a competitive advantage.

By identifying institutions from which they are most-often attracting cardholders, ATM deployers could develop targeted sales campaigns to acquire those customers. For example, one of our financial-institution customers has a particular ATM with high foreign transaction activity. It is now developing a targeted ATM advertising campaign to recruit those foreign cardholders as new customers. This customer's campaign will include special promotional screens to be displayed when one of the targeted cardholders performs a transaction at the ATM, and a bar-coded receipt used to claim an incentive in a branch when an account is opened.

In the third example, Fifth Third mined transaction data to help a customer impacted by Hurricane Katrina. In the first days following the storm, we were able to identify where displaced cardholders were using foreign ATMs for cash withdrawals - thus incurring ATM surcharges - and in just days after the storm, Fifth Third was able to share this data with the customer and work with other financial institutions to waive or reduce surcharges for Hurricane-impacted cardholders. This financial institution and their cardholders saw an immediate, human benefit made possible by data-mining technology.

Data mining makes it possible to answer business questions that in the past we never even dared to ask.

An ATM deployer can analyze EFT network routing and interchange fees to prioritize network routing and increase interchange revenue. The deployer can analyze transaction-denial patterns and trends to quickly address operational issues that might have taken days or even months to appear by just monitoring activity levels. Deployers can monitor surcharge and transaction levels at specific ATMs to optimize income, and they can analyze tradeoffs in surcharge-free networks or targeted surcharge rebate programs to make strategic decisions based on their own transaction data.

Another one of our large customers, which has more than 200 ATMs, was doing an excellent job monitoring transaction trends across its network and at individual ATMs, but the effort was tasking - requiring about two days every month. In 20 minutes, with data mining, this customer set up reports that provided the same information every month, generated automatically.

When the chief executive reviewed the new report, he of course was inspired to ask, "Now, what would happen if we joined a surcharge-free network?"