TL;DR
This post is about one important testing method that should be automated, test data randomization.
You should randomize test data in automated script because this will make your automated test runs slightly more user likely because application that you are testing will be used by different users that will use different data inputs.
Also, it is very usual that application business rules are data driven. Based on data values, feature behaves differently. Using test data randomization, you can discover and test hidden business rules.
Here is my real example. I randomized account number, and soon realized that feature behaves non deterministic. One attribute output values were different from run to run. I could not define the rule.
I asked developers in dedicated slack channel about this behavior, and they explained that this is client business rule. Based on account first digit, created record was labeled differently (output attribute value). Of course, that business rule was not documented in any form.
In my ruby testing scripts, I use Faker gem. Happy test data randomization!
