Making Smart (Enough) Decisions

A conversation with James Taylor, Vice President of Enterprise Decision Management, Fair Isaac

Automating decisions and leveraging business rules to create operational efficiencies is one of the promises of the modern e-commerce age. However, making the right decisions is no simple task. To learn more, we caught up recently with James Taylor of Fair Isaac, a leading provider of enterprise decision management technology and solutions. In addition to serving as Vice President of EDM at Fair Isaac, James has written several books on the topic, including his latest “Smart (Enough) Decisions.” He is also editor of the EDM Blog.

eStara: In your new book, Smart (Enough) Decisions, you cover the topic of how businesses can automate decisions to solve problems and create competitive advantages. In the context of online retailing, what are some examples of how companies are using EDM to improve their interactions with customers?

JT: Online retailing is awash with decisions – some obvious (what cross-sell offer to make to someone who has just purchased something) and some less so (what home page layout to display for a repeat customer). Identifying these decisions and then treating them as an opportunity for improvement and automation gives online retailers more control of their interactions with customers. Instead of one-size-fits-no-one decisions they get mass personalization and instead of deferring decisions to people they can deliver real self-service. A smart enough website, for instance, would display offers or questions that represent the next best interaction with you, allow you to chat or get a call back based on your preferences, question and status, and would apply segmentation and models (for wait time, for example). In addition, it would track what you look at and improve both the content display and offers based on this.

Some specific examples:

• What you display on a home page is a decision. Some online properties are starting to display a standard home page only to someone who is unknown and deciding dynamically what to display to someone who logs in or has a cookie. After all, if you KNOW a customer behaves a certain way, why not use that information to DECIDE what to put on the home page? Why treat everyone the same when you can be more dynamic.

• Loyalty programs, such as the mycoke rewards program, are another great example. Targeting customers with loyalty offers, using offer redemption to drive interactions, identifying the impact of specific transactions on a customer’s loyalty balances (depending on the transaction, date, customer, programs in place, signups, previous activity etc) so as to make rewards more relevant and more rewarding for better customers.

• Continuous pricing, such as Dell Financial Services pricing of credit for its products, replacing static and poorly targeted standard offers. The profit/risk/opportunity tradeoff is different for every customer so online retailers can get closer to the optimum answer by taking more fine grained control of pricing decisions.

• Best next action – what’s the most useful interaction from the retailers perspective. Perhaps it is to make an offer for a specific product, perhaps it is to gather some additional data. Centralizing the decision as to the best next action and then delivering it through whichever channel is next up (website, call center, returns or whatever) requires real decision management.

At lot of this boils down to customer centricity – deciding how to treat each customer each time an opportunity arises.

eStara: Deciding when to offer self-service technology vs. assisted help is a topic most of our readers are curious about.  Based on your experience, do you have any suggestions on how companies can leverage business rules to engage customers with the right form of contact?

JT: Well the first thing I would note is that decisions play in both - empowering staff to act without referring to management and without delay (by automating decisions) makes them a better channel while enabling customers to do more themselves by driving automated decision making into self-service channels will improve that experience also. That said there is also a role for decision management in getting customers the treatment they want.

One of the key tenets of EDM is that of exposing the rules behind decisions to those who should decide what those rules are. Sometimes this means letting customers specify the rules they want applied. Better yet, a retailer could mine data about customers and their behavior and suggest rules based on past behavior to customers – perhaps as a best next action option. For instance, a customer who ALWAYS presses 0 in the call center could be approached (by the next call center rep to speak to them) about having a rule in the system that transferred them to an operator directly as soon as the phone number was identified. Of course, you might make a decision about who to offer this to based not just on behavior but also on value – perhaps profitable customers who always press '0' could be transferred to a special help desk, for instance.

In general, this kind of decision involves not just rules, also predictive analytic models. These might segment customers so that they can be routed appropriately and might predict likely future activities so that these can be made more accessible. Most decisions are a mix of policy, regulation, preference and insight (derived from data about behavior) and this one is no exception.

eStara: How else is EDM helping companies create personalized marketing experiences for online customers?

JT: We are beginning to see multi-channel retailers use EDM to deliver on integration with geographic data. For instance, one retail club chain used analytics to predict cross-sells that would expand the number of categories in which a customer purchased, combined that with rules about local stores and prices and generated personalized offer letters. Replacing the previous direct mail approach with this more decision-centric one resulted in a 2000% increase in responses.

This kind of mass personalization is going to be increasingly important to online retailers. They have so much information – about what their customers buy, what they look at, what they put in the cart but don’t buy – that using it to make sophisticated decisions about every aspect of a customer’s experience is critical to their success. After all, this behavior data is about the only thing they have that no competitor has!

Besides supporting highly personalized and targeted decisions, EDM is increasingly used to deliver real-time decisions so that event-based or trigger-based marketing can be used more successfully. Similarly EDM and the automation of decisions provides a platform for ever-more sophisticated analytics. For instance, as tools come to market allowing the analysis of purchase paths over time (this group of customers tend to buy this product 3-6 weeks after they buy this one unless they buy it in a store), retailers can use this information to make their marketing rules more sophisticated and do so without having to teach everyone how to use the new analytic tool.