Music streaming has colonised the market of recorded music. Recordings of music are transformed from traditional physical artefacts into numerous digital formats, which consequently implies different affordances of how the sound of the music may be mediated. Such digital formats keep evolving which means that how a particular recording sound is a rather undefinable matter over time. Also, cloud storage is replacing traditional archives of recorded music. This entails that sources can be swapped with new versions or taken out of the collections leaving little or no traces behind that it was there before. This digital milieu does, by these features, offer new challenges for research. This paper will present experiences and insights from the method used in my dissertation Liquid Streaming – the Spotify Way To Music where the affordances of musicking brought about by streamed music with the case example of Spotify was explored. For this purpose, digital sources like the Spotify program and sources found on the Internet were used as empirical grounds. The presentation will focus on issues concerning those two major roots by problematizing and conclude how it is possible to use agile software, like the Spotify streaming service program, as well as the Internet as sources of research. One major conclusion from this work is that liquid sources need to be saved as copies and followed longitudinally. My research also showed that sources changed content over time, updated, disappeared, change its resolution e.g. quality. In retrospect, it is clear how digital sources such as programs and sources on the Internet need to be problematized as sources for research in order to bring about trustworthiness to the research.