Meri Gruber, VP Biz Dev here at Decision Management Solutions, had a post a little while ago that she recently tweeted – How many service reps does it take to change a light bulb or lose a customer? In it she makes the point that a company lost her business because the customer service department lacked a clear process and because the service rep had a clear bias – in this case a belief that they could over-charge a woman because they assumed she knew nothing about cars.
In her tweet she said “Would a Decision Management System have tried to charge me $60 to change a light bulb (and lost my business)?” Of course the answer is no.
- First, a Decision Management System would have a consistent set of rules to apply for pricing. Decision Management Systems aren’t biased against particular categories of customers. While most of us try and believe we are unbiased, research has shown again and again that bias happens whether we realize it or not.
- Second, a Decision Management System would have known she was a long standing customer and so, if anything, discounted the price. A Decision Management System would apply business rules and analytics based on the ACTUAL value of the customer to the business rather than be distorted by the person biases of the customer service staff.
- Third, a Decision Management System’s analytics would have identified her as someone likely to be in the market for a new car in the next few years and so have been even more aggressive about discounting (it was a very cheap part/fix and could probably have been offered for cost to build the relationship). This would have valued the profit from a future purchase as well as the margin on the current request. Unlike the customer service reps who were clearly just trying to boost their daily take (and egos).
People have biases and stereotypes. Sometimes these can be effective shorthand but often they are not. Decision Management Systems rely on data (analytics) and explicit policy so they can value customers accurately and treat them right.
The answer to Meri’s question? No, a Decision Management System would not have tried to charge her $60 to change a light bulb. More importantly, it would have recognized her long-term potential value as a customer and taken the opportunity to strengthen the relationship.
What are your people and systems doing?
Comments on this entry are closed.
Good post. Are you saying that decision management systems don’t have bias because having them would would make them explicit, and hence expose the organization to risk (legal,PR)? Technically, I see no reason why a DM system couldn’t have bias. If the system learns optimal choices based on user features, I see no a priori reason why the use of features such as ‘gender’ might not yield different answers.
Matt
I was mostly focused on the rules side of things – the fact that coding bias into rules would be pretty blatant and easy to see/catch/stop. Your point about analytics is an interesting on though – how does one stop bias driven by data analysis?
Two things occur to me. The first is a bit provocative – if it’s real, if the data supports it, is it bias? Well probably not but it could still be illegal.
Secondly I think it reinforces the value of explicable models – understanding and reviewing what your models think is predictive is important and one of the benefits of doing so is that it helps you spot things that might turn out to be illegal “bias”
James