Throughout the development process, we continuously tested the application in different scenarios. In the initial stage, we conducted internal tests by using the application in real-life situations at various locations. Later, we hired freelance promoters who went to different places to calibrate the stores. With regular intensive testing, we discovered various issues with the app and the process, which led to the development of new features in Atlas. As we were expanding to different cities, we had to ensure that the application was not too overwhelming for new users while still introducing new features gradually.

During the testing process, we encountered several issues, such as difficulty in finding items using dropdown lists. After trying different alternatives, we decided to use AutoCompleteTextView available in Android material design which allows users to search and filter the list with keywords. However, we discovered that some users were searching for different keywords and could not find the items. To address this, we implemented predictive search. We attached a list of relevant words with each item so that if they searched for 'shirt', 'Men's clothing store' would filter out.

Another issue we encountered was the time it took to calibrate new zones and groups. Many promoters were entering the wrong addresses, and the process was too lengthy. To speed up the process and make it easier for users, we pre-filled some of the required information using Android Location API when creating a new Zone or Group like names and addresses.

We also encountered an issue where some promoters were using mock locations from home to calibrate, resulting in incorrect calibration data. To address this issue, we implemented a system to ensure that users were calibrating data from the correct places and were not using mock locations. We also implemented other features to eliminate incorrect calibration data.

Furthermore, certain smartphones, mostly Chinese brands like Xiaomi and Oppo, had additional security features that caused problems with location and other sensors. We implemented different workarounds for each brand and their OS versions.

Finally, we faced challenges with using different sensors for collecting data in different scenarios. We had to find alternative ways, such as using Google services for location when GPS was not available.