Skip to main content
Using decisioning to build the bank of the future

I have had some interesting discussions recently around Enterprise Decision Management and banking. These led to the posting"A banking story". It seemed to me, though, that I could articulate more clearly how a The Bank of EDM might act. So, what if your bank...

  • always identified you when you put your card in the ATM, called the call center, handed over a check at the teller
  • remembered your preferences
  • remembered your regular activities and prioritized them
  • accurately predicted your likely behavior/needs
  • applied constraints and circumstances (ATM wait time, call center wait time, teller v personal banker) to its approach
  • used the information you gave them, no matter how you gave it to them
  • and so on...
How might that look? Let's consider some different scenarios to see.

Technorati Tags: , , , , , , , , , , , ,


When you put your card into the ATM it immediately identified you. Based on your expressed preferences and prior behavior it displayed text in your preferred languages and displayed common actions based on your prior ATM usage. If one particular transaction dominated it might say "do you want to perform your usual action or take some other action". If you have a short list, it might list those with an "or something else" option. If there are transactions you perform often using another channel that can also be performed using this particular ATM it might highlight them in a "did you know you can" section. Besides your explicit historical behavior it might try to predict likely behavior using analytic models. For instance people like you who have a larger than usual checking account balance often want to transfer some of it into savings, so it would offer that action. Obviously it would only let you do things you were allowed - it would not offer you more cash than you have or a larger withdrawal that allowed your account type, would not offer actions on account types you don't have and so on.

It might also make a cross-sell or up-sell offer if it predicts that you would have a reasonable chance of accepting an offer and of not being annoyed by it. It would select an offer based on your behavior and segmentation. Obviously it would only do this when its rules said there was likely no line (using recent gaps between customers at this ATM for example), it would not do so if you had said "no offers" and it would offer to follow-up to close any offer you accepted using your preferred channel (call back from sales person, email, link to web form etc).

Is this how your ATM behaves? Should it be?

Call center/Interactive Voice Response (IVR)

When you call, the system first uses your phone number to see if it can identify you and otherwise asks for your account number. As soon as it has identified you, it asks you to confirm and then proceeds, like the ATM, to offer you options based on your prior behavior and predictions about likely behavior. The choices would be driven by common actions you take and uncommon ones based on the behavior patterns of customers like you. The call may be routed to a person immediately based on the banks need to treat you personally or yours if your preferences and actions match. Again, the treatment you get will depend on data-driven segmentation.

When you do talk to someone they are empowered to act on your behalf(see this banking story or this post on self-service in banking). This means they don't have to refer you to someone else, put you on hold while they ask someone and so on. If they cannot do something, or when there is a long wait, the system will use your preferences and models to say when and how you will be contacted.

How does your call center/IVR experience compare to this one?


You're probably getting the picture now. The website remembers you and your actions/preferences. It displays offers or questions that represent the next best interaction with you, allows you to chat or get a call back based on your preferences, question and status and applies segmentation and models (for wait time, for example). In addition it tracks what you look at and improves both the content display and offers based on this. This information, such as your interest in certain products and not others, is fed back into the models about you and your interests.

Like your website?


When you go into the branch, the cross-sell offers displayed to the teller use website and other data about you and lead them in guided interaction so that ever teller acts like an experienced facilitator of banking problems. The systems use the current wait time/queue length to drive speed or quality of interaction so that the teller is efficient when there is a line and engaging when there is time. Anything the teller says respects your channel choice e.g. go see the personal banker over there v we'll email you the information.

How's your branch?

Monthly statement

When you get your monthly statement it repeats the offers generated for you intelligently, asks you to call an 800 number to give feedback on your recent branch experience (which it knows you had and knows is unusual for you) and reminds you of upcoming maturity date on a CD by suggesting a roll-over. In addition it helps prevent identity theft by highlighting critical seeming transactions on the first page of the print out (out of pattern transactions).

and so on...

It can be done. It takes a focus on automating and improving operational decisions, the intelligent use of business rules and predictive analytics and a focus on customers. Enterprise Decision Management, in other words.

related posts