As I said in the first column, this column is going to be focused on decisions and what, exactly, we mean by the word "decision" in Enterprise Decision Management.  To re-cap, here's the definition of EDM that Neil and I use:

Enterprise Decision Management is an approach for automating and improving high-volume operational decisions.  Focusing on operational decisions, it develops decision services using business rules to automate those decisions, adds analytic insight to these services using predictive analytics and allows for the ongoing improvements of decision-making through adaptive control and optimization.

Not one-off decisions, not unstructured decisions, but high-volume, operational decisions.  And not all operational decisions, either.  Suitable decisions tend to have a number of characteristics such as:

• Volume — the number of decisions of a particular type you must make is high.  Volume alone can cause problems or exacerbate another decision problem, such as compliance or risk assessment.
  
  
• Latency — when you can see trouble coming but can't change how you make decisions in time, you might have an operational decision problem.
  
  
• Variability — if the way something is decided in a process or system is highly dependent on outside factors, like competitors or the market, then it could be a candidate thanks to this potential for variability.
  
  
• Compliance — the need to demonstrate compliance of every decision can be a driver.
  
  
• Straight-through processing — a manual review that drags down response time in a process might be hiding a problem-prone operational decision.
  
  
• Managing risk — a risk-centered decision that must be made quickly or in volume might be a good candidate for an operational decision.
  
  
• Unattended — for example, there's no person who can make a decision in transactions on your Web site or at your ATM.
  
  
• Self-service — decision required so that customers can self-serve.
  
  
• Personalized — any time you want to personalize interactions, you're making a decision.

Once you have a decision in mind it is usually pretty easy to ask these questions to see if it makes sense of apply Enterprise Decision Management (EDM) to that particular decision.  However, actually finding all the decisions to which the approach might be applied can be tricky.  While there is a lot of information on this in Smart (Enough) Systems I thought it would be useful to give some quick tips in four areas — two related to finding decisions you already make and two related to decisions you probably don't —manual decisions, contradictory decisions, micro decisions, and hidden decisions.

One of the most valuable sources of decisions for EDM is existing manual decisions.  There are a number of ways to look at manual decisions to find good candidates.  Manual decisions made by supervisors where the request for a decision came from their staff are good candidates because automating them will empower their team to make those decisions directly, reducing referrals and increasing first call resolution, for instance.  Manual decisions in which policy manuals, regulations, or look up tables must constantly be checked are another source of good candidates as are those where more than a small number of factors (3-7) must be considered — people are not good at handling the repetitive consistently nor at managing complex trade-offs.  Manual decisions with lots of "soft" factors are, in contrast, less likely to be good candidates.

Most organizations have some decisions that are contradictory — the website gives one answer while the call center gives another, or where different call center representatives give different answers.  These are good candidates as the management of those decisions can eliminate the inconsistency.  One of my favorite examples of this was an attempt I made to buy tickets for the Chunnel.  Checking from the US we got one price from the website and another from the call center.  Confused, we asked my father in England to check and he got a third price from the website and a fourth from the call center.  After some discussion we decided that he should buy them and he called again, getting a fifth price that was, fortunately, lower than the other four.  If your organization has decisions it does not seem able to deliver consistently, look into those.

The third area is one of those related to decisions you probably don't see today.  Micro decisions are the little decisions that contribute to an overall decision or action — the decisions made in each individual transaction.  For instance, a direct mail campaign could be considered to have 4 or 5 decisions — what message, what segment of the customer base, what format, what offer, and so on.  If the direct mail piece went out to 100,000 people, however, you have 100,000 micro decisions.  You decided in each of 100,000 cases to send this particular letter to this particular person.  If you thought of that decision — what to send to this specific person — you would have new opportunities to micro target and personalize that decision.

The final category is that of hidden decisions.  These are decisions that never get taken.  For instance, my ATM (once I have entered my PIN) always shows me the same options.  It does this even though I never do anything except take out cash or pay in checks.  It displays options I can use and others I cannot (because I don't have the right accounts) and it fails completely to identify me as a gold customer, though I am one.  The decision as to what to display to me was never thought of as a decision — it's just hidden in the code to run the ATM.  Kiosks, websites, and devices are often full of these kinds of decisions.

These categories are not mutually exclusive — a manual decision may well be contradictory and might be replaced with a micro decision, for instance — but they should give you some ideas when you are looking for a decision or two to automate.

Next time, data mining and the use of data to find and improve rules.

Don't forget you can use RSS or e-mail to subscribe to the Smart (enough) Systems blog.

# # #