During the last decade social media have greatly flourished, reaching rapidly the amazing figures of today. According to the Search Engine Journal (http://www.searchenginejournal.com/25-insane-social-media-facts/79645/): (a) currently 684,478 pieces of content are shared on Facebook every minute, (b) people are spending 1 out of every 7 minutes on Facebook when online, (c) 93% of marketers are using social media, however, only 9% of marketing companies have full-time bloggers and (d) around 46% of web users will look towards social media when making a purchase. It is obvious that businesses are tapping into social media, since they find them as a rich source of information and a business execution platform for product design and innovation, consumer and stakeholder relations management, and marketing. For this reason it is very useful to evaluate the impact of each posted advertisement. Towards this direction several supervised works have been presented in literature mainly focusing on traditional media. However, the impact of advertisements on new media (such as social networks, blogs etc.) has not been studied thoroughly yet. Additionally unsupervised impact evaluation is a very challenging problem. In this paper a novel unsupervised social computing approach is proposed that effectively performs both on open social media (twitter, blogs, microblogs etc) and on rule-stringent media (e.g. Facebook, LinkedIn etc). Our scheme algorithmically estimates the importance of each advertisement by considering both explicit interactions between advertisements and social media users and users’ popularity. The proposed method operates without human intervention and training and it is applied on real content posted on social media. Experimental results provide an insight of the performance of our system and specific areas are detected for future research.