The security threat of data leakage is increasing, and perimeter protection mechanisms such as firewalls are no longer sufficient to safeguard sensitive information. Data loss prevention (DLP) solutions are promising, but the lack of transparency in this area makes selecting the right solution challenging. This paper provides a systematic evaluation of content-based DLP solutions to determine their effectiveness in preventing data leakage in web traffic. The study focuses on HTTP, the dominant internet protocol, and reveals that while DLP solutions protect against accidental data loss, attackers can still bypass them. To address this challenge, The research provides an artificial neural network-based content classification technique based on MLP architecture and an n-gram TFIDF feature descriptor to detect and protect sensitive information of a well-known TI business. When compared to existing solutions, the proposed technique is found to be significantly superior at minimizing insider threats and preventing data leaks.