Consider an eCommerce system where customers can buy stuff. Part of handling commerce is handling faults. Those range from “I bought the wrong thing” to “my kid just bought a Ferrari”. Any such system will need some mechanism to handle fixing those faults.
The simplest option we have is the notion of refunds. “You bought by mistake, we can undo that”.
In many systems, the question is then “how do we manage the process of refunds”? You can do something like this:
So a customer requests a refund, it is processed by the Help Desk and is sent for approval by Finance, who is then consulting Fraud and then get sign off by the vice –CFO.
There are about 12 refunds a quarter, however. Just the task of writing down the rules for processing refunds costs more than that.
Instead, a refund policy can state that anyone can request a refund within a certain time frame. At which point, the act of processing a refund becomes:
Is there a potential for abuse? Probably, but it is going to be caught naturally as we see the number of refunds spike over historical levels. We don’t need to do anything.
In fact, the whole idea relies on two important assumptions:
- There is a human in the loop
- They are qualified to make decisions and relied upon to try to do the right thing
Trying to create a process to handle this is a bad idea if the number of refunds is negligible. It costs too much, and making refunds easy is actually a goal (since that increases trust in the company as a whole).