Text classification, is a very popular problem with various applications in natural language processing (NLP). One of the core tasks performed in text classification is assigning labels or tags to units in the text data such as sentences, paragraphs, and documents by exploring the relation between words or even characters. There are many applications derived from text classification, namely Sentiment Analysis, Topic Classification, Spam Detection, Document Classification, and so on. The main object of analyzing is text data. It can come from various sources like a newspaper, a document, some text messages that people use on daily basis. Naturally, as one of the most important forms of communication, text is an extremely rich source of data. However, due to its unstructured nature and highly dependence on the context of use, extracting insights from text can be very challenging and time-consuming. This study focuses on exploring the data and forming a classification model on some of the gaming application test sets. We approach the problem using some basic text analysis methods and performing text classification by applying a Deep Learning method—the Convolutional Neural Network model. The dataset is collected from the handwritten test sets for various in-game content by the Quality Assurance Engineers. The main label to be classified is the Priority of the test cases on a whole test set, and eventually, the priority will be used to choose which test cases fall into the Regression Test set, specifically four types of Priority from highest to lowest label. Finally, we provide an analysis of the performance of deep learning models based on the evaluation metrics as well as comparing it with a self-built traditional Machine Learning model using Logistic Regression and testing against real test case input. From that, we expect to learn to improve the deep learning model and discuss the possible future directions.