Are you ready for decision management? Part 4 - James Taylor's Decision Management

« Are you ready for decision management? Part 3 | Main | Are you ready for decision management? Part 5 »

November 30, 2006
Are you ready for decision management? Part 4

How do you tell if your organization is ready for decision management? If not, what can you do to start laying the groundwork?

Fourth, Organizational Change

Automation of operational decisions can cause a fair degree of organizational change. Some roles will change - they will go from making large numbers of simple decisions to spending their time considering the overall patterns of decisions. Thus underwriters might go from spending most of their time considering individual policies to spending most of their time doing analysis of the relative success of agents or areas of the state. Advertising managers might spend their time on account management not on calculating ad prices. This change will mean that some people will go from being good at their job to being less good, and vice versa. This new focus will need to be articulated to staff and the implications managed. If staff reductions are not planned then reassurance will be needed that the computer is not going to replace people. If staff reductions are expected then this will have to be managed. Organizations that have a hard time managing this kind of change may want to start with projects that automate decisions not taken at all in the past (adding a cross-sell engine for example) or that are already automated poorly (a generic cross-sell embedded in the website for example).

Where staff are relying on the decisions made by a decision service and are responsible for passing it on to customers - such as a call center representative using a pricing engine - there may well be a reluctance to "trust" the computer's judgment. This is especially true if the staff have bonuses or similar tied to the effectiveness of the decision. One way to manage this is to implement the new decision service incrementally and make it available only to a small group of staff (those more willing to change) to show it works. Retrospective analysis can also be used to show how much more successful the organization would have been if they had used the new service in place of the old judgmental process. If this can be tied to statements like "you would have made X more money" then so much the better. Nothing succeeds like success so try and demonstrate success with a few or that the new approach would have been more successful than the existing one. Be prepared for initial concern no matter what you do - there will be a spike in complaints/appeals etc initially but if the decisions are sound it will wear off.

Customers getting automated decisions may be glad to get more done in a self-service way but will likely also feel aggrieved if they don't get what they want "because the computer said so". Managing this requires decision services that explain their decisions, at least to staff if not to customers. Decisions that say "no but" rather than just "no" will also do better. For instance, "you cannot have that product but you can have this one". Be sure to have legal review any decisions that are regulated or that have historically generated lawsuits over bias on the part of staff. Making sure the rules and analytics you use are defensible in advance will be easier than running around in the face of lawsuits. Again, introducing decisions incrementally and with explanation, especially when the decision is important to a customer (loan approval is, cross-sell is not) will be important and should be planned for.

Finally beware of counter productive incentives. Sometimes the sophistication possible in an decision management solution will overwhelm the performance and reward structure. For example, marketing staff may have objectives for both volume and response rate. A decision engine that dramatically improves the response rate may reduce the volume of offers mailed and this may cause people to miss their bonuses which will, in turn, make them unwilling to use the decision service. Where decision services are going to affect the behavior of staff, play out the impact you hope for of the decision service on their behavior and see if it increases or decreases their pay. If they lose out, but the decision service is making decisions that are better for the company, make sure this is addressed as part of the rollout.

Technorati Tags: business agility, business rules, decision automation, EDM, Enterprise Decision Management, innovation, readiness, organizational change