Ever increasing fame and obsession for social networks has also coxswained a dramatic increase in the presence of malicious activities. As a result, various researchers have proposed different features and techniques to detect and reduce this menace. This paper presents an expository study of various state-of-the-art techniques to detect two most interlinked apprehensive problems on social networks namely, spam detection and detection-cum-analysis of compromised accounts. It is evident from the ongoing statistics that despite profuse awareness and some anti-spam policies and techniques being developed, to everyone’s surprise, the severity of spam has only increased. Moreover, with the growing smartness of spammers, existing techniques get bypassed and new features and techniques continuously keep on evolving. Therefore, in continuation to the ongoing research, a study comprising of a comprehensive analysis of different works is also required from time to time. The growing inclination of spammers to compromise the legitimate accounts has evolved as an evasive and more beneficial way to spread spam. Therefore, it has become highly relevant to review the techniques related to the detection of compromised accounts so as to track the spammers adhering to this behavior. In this work, we have performed a qualitative analysis of each paper discussing its pros and cons. In both the domains, the detection approaches have been placed under different categories and have been thoroughly reviewed stating their applicability. In conclusion, the paper carries a discussion of various gaps prevalent in the existing approaches and the corresponding actions to be taken to address them, providing a strong foundation for future researches to be carried out in this domain.