Research libraries are putting in a lot of effort to make their collections available to researchers, students, and the general public. Through digitisation, crowdsourcing of metadata, online exhibitions, and initiatives focused on creativity and remixing, the cultural heritage of Europe is made available and put into a contemporary spotlight, whether it is through artificial intelligence image analysis or quarantine at-home remakes of famous paintings. While making collections accessible in online repositories and search portals is important, challenges remain in increasing interaction with the digitised collections. One way of doing this has been to use crowdsourcing. This co-producing method increases exploration and interaction with collections, but yields varying results in terms of coverage and quality. Crowdsourcing is continuously explored and evaluated in many fields of science, and one identified success factor is community building. As a librarian I wanted to explore how we as a research library could facilitate community building in crowdsourcing while also experimenting with novel ways to enhance collection metadata, teach data literacy, and support research projects. In terms of digital humanities, it is important to de-mystify data and coding, so that digital humanities researchers can have an audience who understands their methods (even if they cannot apply them). The (albeit constantly renegotiated) gap between digital humanities and humanities needs to be bridged, and libraries teaching data literacy is a partial solution to this issue. In this case study, I sum up the experiences from eight cultural heritage hackathons I organised in 2018-2020. The hackathons were connected to five different themes, related either to humanities research projects or to specific parts of the library collections. While a “traditional” hackathon focus on writing code together, the cultural heritage hackathons were focused on performing a digital task together. The term “hackathon” was chosen to de-mystify data as a concept and to bring different user groups together. (In the physical events we also fully embraced the hackathon trope of soft drinks, pizza, and coffee.) In the hackathons, participants were tasked with adding or editing data (e.g. OCR corrections, transcriptions, geotagging, image tagging). Difficulties, clarifications and common practices (e.g. deciding transcription conventions) were discussed and solved among the participants, creating an atmosphere of trust and co-ownership. The hackathons varied in type of audience (some were open for all, others only for invited participants (so-called “expert sourcing”)), medium (online, hybrid, physical events) and tools (spreadsheets, dedicated platforms, or built-in solutions). The multipronged approach aimed to cater to several stakeholders at once and partially succeeded with this goal. However, some issues (e.g sustainability and completion rate) still need to be solved. The experiences from the hackathons have resulted in best practices and recommendations for libraries who wish to explore this methodology further. An important outcome is knowledge of the role libraries can play as an intermediary between researchers, the general public, and librarians, and how this can result in mutually beneficial practices between all parties, thus enhancing data literacy skills, supporting research, and increasing library outreach.