Skin cancer represents a prevalent cancer type, and early detection significantly impacts patient outcomes. A concerning rise in the incidence of various skin diseases is found nowadays. This alarming movement has not only affected the overall health and well-being of people but has also led to a notable rise in the development of skin cancer cases worldwide. This paper focuses on the development of a skin cancer detection system using a deep learning-based framework. It has been observed that deep learning outperforms conventional machine learning methods to a great extent. Among different deep learning methods, the Convolutional Neural Network (CNN) is very popular since it can handle spatial data well. The objective of this study is to leverage the performance of detecting skin cancer among the pool of skin lesions as benign or malignant by providing an automated tool to assist dermatologists in diagnosing the disease correctly in no time. Here, we have developed a CNN model to classify the skin lesions as benign or malignant. The performance of the model was evaluated using different performance measuring parameters such as accuracy, precision, recall, and F1-score. The key findings of this paper demonstrate the effectiveness of CNNs in skin cancer detection. The developed model achieved a satisfactory performance in classifying skin lesions, outperforming existing methods. The interpretability of the model was also explored, shedding light on the learned features and contributing to its potential as a diagnostic aid. Overall, this paper contributes to the advancement of computer-aided diagnosis systems for skin cancer. The developed CNN-based model shows promise in assisting dermatologists with accurate and efficient diagnoses.