Effective detection and classification of malicious URLs are critical components of safeguarding against cyber threats in today’s digital landscape. Malicious URLs serve as gateways for various cyber-attacks, including malware distribution, phishing attempts, defacement activities, and spam propagation. Detecting these malicious URLs requires advanced techniques that can analyze and distinguish between benign and harmful web entities. Machine learning (ML) algorithms have emerged as powerful tools in this endeavor, leveraging diverse feature categories and sophisticated classifiers to differentiate between legitimate and harmful URLs. In our study, we evaluated the performance of ML algorithms, with Random Forest achieving notable accuracy rates of 96.83% for binary classification and 92.13% for multi-classification, outperforming other classifiers. Our investigation spans Lexical, Content Analysis, Identity Verification, Identity Similarity, Visual Similarity, Behavioral Analysis, Entropy Analysis, and Geographic Analysis features. Our research seeks to provide comprehensive insights into the methodologies and approaches employed in malicious URL detection. This research adopts a systematic, manual process to identify and rank critical URL characteristics, ensuring relevance and interpretability. The findings provide comprehensive insights into malicious URL detection methodologies, highlighting effective strategies for both binary and multi-class classification tasks.