Combining exploratory testing and automation

The client

Morphean offers a video surveillance platform responsible for data gathering for store optimization and video analysis. Certain events – specific types of movement – will trigger the platform to send out notifications. The platform offers access control through doors using saved presets of PIN or cards credential types saved for a person. Actions done using those credentials are audited in an access log for reviews.

Our testing approach

Given the high variability of the devices and patterns that customers could have, we opted to use a combination of exploratory testing, alongside existing test cases and test automation, at both UI and API level.

Exploratory testing

Traffic around a surveillance camera can trigger a series of actions at unpredictable intervals. Access logs can have different credential types that require physical interactions on an access control device based on the configurations done for a group of doors. Using exploratory testing we could see how these devices react in a real environment where unpredictable situations can arise.

A lot of issues we discovered were specific to one device, a version of the device’s app, or a device’s firmware. Our approach enabled us to set up the right configurations and choose the right tests to run. Through exploratory testing, we observed new feature changes and other relevant information besides the usual testing flow. This helped us prepare the acceptance criteria for development so they could know how the app should look and function.

Test automation

Implementing an automation testing strategy besides exploratory testing, was a response to our client’s needs for fast feedback and validation of release.

Due to stories being split into backend and frontend, API testing helped with validating the feature in the development stages of the backend. With this approach, we managed to test authentication permissions with different user roles, error handling for incorrect payload, and what responses we get for different payloads.

We also realized that by automating some of the existing test cases on this project, we would have more time to investigate the tests that failed pointing out areas of greater risk on which more exploratory testing was required. So we continued onto automating the smoke test cycles and the repetitive regressions in order to validate a release automatically.