Split Testing Processes for Performance

Marketing gurus have long used split testing as an iterative process to find the best performing adverts in direct mail marketing. It is a simple process where 2 different adverts are created, sent out, and the responses to each ad are recorded. It is generally referred to as A/B split testing. The conversion rate of ad A is measured against ad B and the worst performing advert is then dropped. A new ad is created and put through the same process against the remaining (better performing) ad from the previous round of A/B split testing. Simple but effective. Many companies now do it with their Google Adwords. So what does split testing have to do with business process and performance management. Everything…

Why don’t we split test processes for performance optimisation? Why don’t we create two streams of a process and use some randomisation algorithm to send a particular transaction down a specific process stream? We can then record and measure the performance of each stream based on pre-defined process performance indicators to the process analytics module of the BPMS and determine the best performing process. Rinse and repeat.

This iterative process will provide an alternative to modelling and simulation techniques that many process analysts employ to try and improve process performance. It provides a practical approach that proved successful in other areas such marketing and it is an evolutionary approach where the best performing processes survive. Many organisations implement a business process without ever re-visiting it. No one knows where to start looking for changes. This split testing process will take care of that.

A word of caution thought. Make sure that the BPMS solution can support this type of versioning of processes and that it doesn’t impact negatively on the quality of information on the user interface and that audit trails and worklists maintain their integrity. Processes need to be designed for split testing.

The possible benefits of A/B split testing for processes can be significant in organisations where process improvement plays a significant role. It can possibly save jobs and improve bottom lines in the current economic circumstances. It is worth a try.