Uncertainty about a company’s finances is a problem because, due to information differences in the financial sector, stakeholders usually do not know how well a company is doing until crisis or financial trouble begins to develop. We need to create effective systems for predicting failures since financial distress endangers both the operations of a company and the interests of shareholders, might affect the country’s economy and could negatively impact the community. To achieve better and stable results, this study uses both DNN and CNN as major deep learning tools. Also, the chi-square automatic interaction detection (CHAID) method is used to select the most important variables that help predict outcomes. The study looks at data collected from the websites of about 330 companies from the Bombay Stock Exchange (BSE) in India, including 80 that faced financial crisis and 250 that did not. The results indicate that using a combination of CHAID and CNN makes it possible for the model to make the most accurate predictions of financial distress compared to the other models, with a result of 94.23% accuracy. Among the examined models, this method gives the lowest probability for making both the Type I and Type II errors.