Originally Published: April 17, 2013 – Updated: June 2020
I have been involved in many different kinds of projects in my six years of implementing supply chain systems, and very few of them invested in automated testing. While there is a lot to discuss and unravel in this topic, I will focus on my personal experience from a managerial point of view.
To Automate or To Not
Think about automating testing as a tool that hammers the system with real-life test cases without the need for a human to run it. The primary purpose of such an exercise is to validate the functionality and stress test the system in an attempt to fix business logic and increase the performance. Looking at this from a manager's perspective, what would I need to worry about considering the great benefits of a successful automated test?
Focus on the main scenario, is it worth it to automate the testing of some value-added service in warehousing? In my years of implementation, I noticed that due to the nature of various industries and their lines of business, some customers invested in automated testing while others did not. I recommend customers with more than 1DC, and who are processing an upward of 10K order lines a day to think about automated testing.
Within the four walls of a warehouse, the scenarios are numerous, so focus on your critical path as time. Time is a highly valuable consideration and a commonly underestimated factor in many projects. Allocation, release, host interfaces, and MHE (material handling equipment) are significant candidates for automation testing.
Preparing a Data Set
When preparing to test, first prepare your data set needed. It may take a couple of iterations before you get a successful start to finish test. For example, during one of my projects, we ran an automated test for an automatic storage and retrieval system (ASRS).
By doing this, it meant populating thousands of locations with various-sized inventory and covering various pull and place scenarios. Initially, this data setup took a considerable amount of time. It became much easier during consecutive runs. The initial run required lots of effort to end up with a smooth, non-interrupted, and successful test.
Involve Operational Staff
Lastly, human error during a go-live is often hard to predict. In all my go-lives, there have been surprises. Users interact with the system using an RF device and the GUI screens. Most human errors are harmless. However, the remaining few that are harmful end up costing a lot. So, engage your operational guys and let them review the tests first.