Adventures in the Customer Data Mine

I’m just minding my own business teaching a .NET development class down at the Learning Annex and I get this call from a puzzled banker. He says he’s got a data mine that he’s extracted some customer activity and transaction data out of for fraud monitoring and BSA purposes and he’s got the scars to prove it. Now he’s hearing rumors there might actually be some treasure in there somewhere. He was told I was the best there was at finding hidden gems and looked me up. Well, you know Linden Jones is always up for a little adventure so I decide to take a few days off, hop in a plane, and head to Cairo. Yep, haven’t hit South Georgia in years.

So I get down there and find my new banker friend Eugene buried under a pile of acronyms in his office … BSA, KYC, AML, MSB, CIP, PEP … yikes! “Eugene, what the heck happened?” I asked as I pulled him out of the TLA⁽¹⁾ rubble. “Well, our data warehouse project’s been going on for a couple of years, Lindy. We’ve been working with the auditors, operations and BSA groups adding feeds, tables and data elements until we had something we could do some fraud reporting against and feed our fancy new BSA/AML tools from. There’s a body count from all the iterations of development we did on that warehouse.

“Now some whippersnapper in Marketing is spouting off that there might be something more valuable hidden in the data mine. Can you believe it?” the banker asked.

“I wouldn’t be here if I didn’t, my friend,” I replied. “Let’s head on into the mine.”

As we enter the dark corridors of the warehouse we don’t have good design documents to use as a map, so we rely on a database schema and some artifacts we picked up from the interfaces from the source systems. Soon, we reach a cavern with some interesting relics of customer and account information including:

• CIF information
• Account balance history
• Transaction history, summarized by channel and maybe even some detail

After a few months of accumulation, that stuff ought to be enough to risk-rate customers, do some fraud detection, and flag some money laundering suspects. Know what else you can do with that kind of data? Generate some retention and cross-sell leads … some real gems.

“Guess what—your marketing guy was right. Ever heard of event-based marketing or predictive analytics, Eugene?” I queried. “Nope,” says Eugene. So I start my explanation. “Well, those kinds of tools will let you take some of that data you’re using for defensive, risk management purposes and also let you use it to retain customers as well as cross-sell to them.”

The nifty thing about these systems, I tell him, is that the leads are more effective than the lists you’re probably sending to your branches today. “How are your branches doing working those closed account winback reports and making calls on significant balance changes?” I ask.

“Not well, Lindy,” was the unsurprising response. “By the time the account’s closed and they get the report it could be a couple of weeks before a call is made, too late to get the account back. And the significant balance change report is giving us way too many false positives, so a lot of conscientious objectors have decided to stop using it. But I wouldn’t know that from reports because the reporting is a disaster, you know, all legal pad and Excel-based.”

“Well, Eugene, you’ve got the data to do something about that,” I say. “There are solutions on the market today that will address the problems you just laid out. Take your closed account report—wouldn’t you rather be reaching out to the customer while the account is still open? A good tool will look at account behavior and tell you when a customer’s activity is waning—perhaps she stops using her debit card or bill payment, a sign the checking account might be moving. By getting that lead sooner, the day the activity decrease is recognized instead of a week after the account closes, you’re more likely to keep the account.”

How about that significant balance change report, I ask him? By looking at account balance history and determining what is significant from that history, a good analytics tool realizes that $25,000 isn’t a significant change for everybody and adjusts its threshold at the individual account level. That eliminates a lot of the false positive leads, unnecessary calls, and frustrated customers and customer service reps.

The reporting on what happened with the leads is automatic, I explain, from whether the calls were made to whether the account was retained. You can run onboarding programs with these systems and the fancier analytics solutions can even provide real-time leads back to your various delivery channels for presentment. “Pretty cool, huh?”

“Lindy, you’re a genius—let’s mine some leads and make some calls today!” That’s when the giant rock dislodged and started rolling through the mine at us.

“Oh great, Eugene, now you’ve ticked off the analytics gods,” I holler above the roar. “Think before you do something rash like that! Before you make call number one you’ve got to get some things in place!”

As we’re running out of the data mine I explain the things he’ll need to do before he gets his new event-based marketing program up and running.

• If your marketing area doesn’t have the expertise, partner with someone who understands the behaviors that lead to account attrition and cross-sell opportunities.
• Modify extracts as necessary to make any additional data required available with the desired frequency.
• Ensure the marketing department has a system that allows it (not I.T.) to edit the criteria for creating events (generating leads) itself and which can create campaigns to test refined criteria before rolling it out throughout the institution.
• Don’t overwhelm branches with a volume of leads they can’t possibly work—they need to act quickly on them, as the shelf-life of some of these leads may be just a couple of days. Rank the types of events by payoff and get the events with the most opportunity in the front of the call queue.
• Provide solid training and scripts to your customer service reps, so customers don’t feel like they are being spied on when they receive a call. Don’t forget relevant privacy and opt-out provisions, too.
• Follow-up on goals set for working the leads; the system has little value if the leads aren’t being worked consistently.

As I make my last point, the boulder crumbled behind us. “Well, Lindy, looks like we’ve got a bit more work to do before we make those first calls,” Eugene quipped. “But boy am I excited about the treasure you found in that dingy old mine—thanks!”

My adventure was done but I’m up for more, so call me if you need some help finding those hidden gems.

⁽¹⁾ Three Letter Acronym

-QS