I was traveling in South Africa last week (keynoting BI 2010) and my favorite online payment system demonstrated not once but twice, why business rules are so valuable in analytic decision making. First their analytics triggered a fraud alert – presumably based on patterns of problems from South African IP addresses. As I was trying to make a payment to a vendor I have used before, a simple rule could and should have overridden this (if the payment is to someone who has been paid before without problems then it is not likely to suddenly become fraudulent). But it did not and I had to go through the unblock account process.
Sadly the unblock account process assumed you were at home. Of course, I was not, and there was no way to unblock the account. This is a problem as most fraud alerts are false positives (that’s just how it works) so the assumption that I am traveling would have been more reasonable than that I was at home. Some rule to say that, if it was me and not a fraudster, that I was likely to be away from home and so could not answer my home phone would have been helpful. As it was I had to leave the account suspended until I got home.
The account is now unblocked but, trying to be helpful, I sent them a note explaining the situation. Sadly their use of text analytics was also rules-free. Picking up the words in my email about suspended account processes it simply sent me a note about how to unblock my account. Again, a simple rule would have discovered that I had, in fact, already unblocked my account and so probably did not need help on that topic. This would have allowed them to route it to someone who could have answered my email instead of merely irritating me with an automated response.
Analytics can detect fraud and give you a good sense of what an email says. But explicit business rules can map that insight to current business conditions and make sure you action makes sense. Use them together and make better decisions.