About a year ago I started harping on the test automation bandwagon, and had introduced basic smoketests & New Relic Synthetics 'scripted browser' monitors (both Selenium-based). In the spring of 2016 I used the Pluralsight course 'https://app.pluralsight.com/library/courses/automated-testing-framework-selenium' to help build our inaugural UI test framework. I wrote out a number of smoketests and 'hey this bit us on the last deploy, let's write a test for it', and the idea quickly proved itself. Very shortly afterward we hired a QA guy specifically for automation experience, and now...
- He took the framework to a whole new level, and it's now the defacto way forward for QA
- More QA folk were hired, and they are now learning to use the framework to write mundane tests
- Some QA folk initially were 100% in the camp of 'QAs do not write code, ever' - and now they are pushing themselves forward - learning not just our UI framework, but learning API testing as well!
- The initial QA automation hire (who continues to be awesome and an inspiration) has also written an API test framework in a similar vein, and we're now starting to adopt that as 'how we test APIs'
- The future is very exciting for us, and more and more 'high business value' things are coming down the line...
This is a very brief success story, but wanted to share for those stuck in 'this is impossible'-land.
Our path to software quality started only a year ago with...
- Zero QA staff, little monitoring
- Some QA staff, but not interested in automation, very basic monitoring
- Semi-automated smoketests and scripted monitors
- A UI test automation framework borne from a month of evenings/weekends and a Pluralsight course (zero experience)
- A functional and useful framework taken to the next level by a QA automation hire
- More and more interest in QA to join the automation crew
- API testing using SoapUI
- API test framework that will leverage Postman code generation
- Investigation into Cucumber/SpecFlow...
You won't know until you try...
P.S. I should note that much of this is 'after the fact' software quality. In hindsight it might have been more effective to push harder on centralized logging/metrics (which is now also a thing, and super amazing). You should also note that people made this successful. :)