Sliding-scale not binary decisioning

Back in July when I was interviewed for DM Review I discussed sliding scale decisioning briefly and I realized today that I have not expanded on this. In the article I discussed how I see a move from simplistic to complex decision making and that for companies that also need speed of decision making then automation of these decisions is critical.

When most people think of decisions they think of yes/no types of decisions - binary outcomes essentially.

• I have applied for a loan, do I get it?
• Should I try to retain this customer?
• Will this package arrive on time?
• I want this cell phone and plan, do I get it?

In reality, however, there is typically a much more complex decision-making process.

For example:

• What kind of loans and at what rates am I eligible and which best meets my needs?
• What offer maximizes my chance of retaining this customer at a price that makes sense given the predicted profitability of this customer?
• What is the likelihood that this package will arrive in any given delivery window and what options do I have to change this?
• Given my credit history and other characteristics, for which phones and plans am I eligible?

These more sophisticated decisions are much more "sliding scale" decisions in that various finely graded outcomes are possible for each transaction. Taking this approach means that each decision can be evaluated so as to:

• Manage risk
  Typically by using analytic approaches to develop models that predict risk and then offsetting the risk against the potential benefit of each point on the sliding scale.
• Manage opportunity
  Allow for the fact that using resources to go after this opportunity mean they cannot be used to go after that one and build these kinds of tradeoffs into the decision process.
• Focus on individual responses
  Personalize or customize each decision so as to more accurately target that customer. The more segments or gradations of decision available, the more targeted the offer or action.

Much of the blog is really about this and there are some other posts on topics that came up in the interview:

• What is agility and why would you care?
• Managing decisions as a corporate asset
• Transaction-centric decisioning
• Strategic alignment for operational decisions
• Decision-making in Service Oriented Architecture

Technorati Tags: analytics, decision automation, EDM, Enterprise Decision Management, personalization, predictive analytics, risk, service, SOA, DM Review

Comments

[Colin Henderson]

While I agree on the concept, as a banker, I worry how bankers make a literal interpret of this capability. One of the disconnects that banks and their customers have is centred on credit decisions that say, approved, with the following conditions. Those conditions can be as simple as "provide proof of income". The problem is that its easy for the lender to ask for proof of income every time, but do they really need it?

Decisioning needs to take some extremely basic physical aspects into play in order to provide high levels of customer service, but these must be hard to automate, as in " proof of income has been provided once in the last "x" years, therefore is not required".

These are very low level decision points, but are also the ones that create the most custeomer service disconnect.