Research problem: The purpose of the study was to explore how a boundary spanner can successfully facilitate knowledge sharing across functional and geographical boundaries. The main research questions are: (1) Does matching the complexity of knowledge boundary with the knowledge-sharing process lead to successful knowledge sharing? and (2) What are the key factors that influence a boundary spanner when deciding how to facilitate the knowledge sharing across functional and location boundaries? Literature review: The purpose of the literature review was to better understand the existing knowledge-sharing frameworks. Finding no framework that can directly address the research problem, the researchers decided to build on the closest one which is a framework for knowledge sharing across functional boundaries. By not taking into consideration knowledge sharing across location boundaries, the framework assumes collocated and synchronous knowledge-sharing interaction. To understand the potential supporting media for knowledge-sharing interaction across functional and geographical boundaries, the researchers consulted the Media Naturalness Theory. Media naturalness is the ability of the media to support a sense of collocated and synchronous interaction. Methodology: The researchers conducted a qualitative exploratory case study in the IT department of a Fortune 500 multinational finance company. Researchers selected a boundary spanner and observed her facilitation of knowledge-sharing interactions for four months. A total of 78 knowledge-sharing interaction logs were collected during the period of observation from five data sources: wiki, email, instant messaging, teleconference, and face-to-face interactions. Data analysis was carried out through template coding. Results and discussion: The researchers found that matching the knowledge boundary with the knowledge-sharing process is an essential yet insufficient condition for successful knowledge sharing. A boundary spanne- should also pay attention to the boundary objects and media used to support the knowledge-sharing interaction. Spatial dispersion and knowledge commonality between the source/recipient and boundary spanner affected the media selection which, in turn, influenced the selection of the boundary objects. The implication of the study is that there are three important factors that the boundary spanner should consider when deciding how to facilitate knowledge sharing (i.e., knowledge boundary, spatial dispersion, and knowledge commonality). The main limitations of the study were the relatively short observation period of the knowledge-sharing interactions via a boundary spanner. Future research should quantitatively validate the proposed optimal knowledge-sharing designs to test the generalizability of the findings with a survey and profile deviation analysis.