Discovery and detection of different social behaviors, such as influence in on-line social networks, have drawn much focus in the current research. While there are many methods tackling the issue of influence evaluation, most of them base on the underline assumption that a large audience is indispensable for an influencer to have much impact. However, in many cases, users with smaller but highly involved audience still are highly impactful. In this work, we target a novel problem of finding micro-influencers - exactly those users that have much influence on others despite of a limited range of followers. Therefore, we propose a new concept of micro-influencers in the context of Social Network Analysis, define the notion and present a flexible method aiming to discover them. The approach is tested on two real-world datasets of Facebook and Pinterest. The established results are promising and demonstrate the usefulness of the micro-influencer-oriented approach for potential applications