Cloud computing is a new paradigm that provides end users with a secure, personalized, dynamic computing environment with guaranteed service quality. One popular solution is Google cloud firestore, a global-scale not only structured query language (NoSQL) document database for mobile and web apps. Recent research on cloud-based NoSQL databases often discusses the difference between them and SQL databases and their performance. However, using cloud-based NoSQL databases such as firestore is tricky without any scientific comparison methodology, and it needs analysis of how its particular systems work. This study aims to discover what is the best design that could be implemented to optimize data read cost, response size, and time regarding the cloud firestore database. In this study, we develop a grade point average (GPA)-report mocking application to assess data read based on our institution’s needs. This application consists of three functions. Add the graduated GPA and students’ names, and view the ten highest GPAs, GPA average, and total graduated students. The finding indicates that aggregating data on the client side or utilizing the Google cloud function trigger, then updating aggregation data in one transaction significantly reduces document read count (cost), response size, and time.