Data Cleaning: Fringe Case Data Anomalies

Data anomalies can be small (less than 5% of your total data set) or large (50%+), depending upon many factors.   One factor I will explore today are business operations “fringe” cases.  
   
What is a fringe case?  Let me explain.

A fringe case is a business process (whether planned or unplanned) by your customer service/operations teams to “make your system work” to fulfill a promise or fix an issue with your customers.    These typically occur when a business process is broken, or where there is a limitation with your existing system to “do right” by the customer.   

Fringe cases can occur when there is a change in one of the following scenarios:

1. A technology changes or upgrades occur in your legacy system

2. A policy change with how you do business with your customers

3. An internal process change that impacts how your employees engage with your customers or systems to support your customers

In each of these cases, human intervention occurs within systems to “get the job done”.   At face value, you should be proud that your frontline teams are being creative problem solvers on behalf of the business.   However, there are downstream impacts to your data, and your reporting.   

Fringe cases can impact your underlying data, how it is processed, and how it is then represented in your reports and analysis.   From my experience, my teams often learned of technology/systems issues when we saw changes in our daily reporting on performance.   However, if you do not have a mature reporting ecosystem, you may not be aware of these problems.