Cross-agency data-sharing collaborations have proven crucial to ensure effective usage of often limited funds. Through, delivering these resources to the right people in the right circumstances at the right time. Despite the cumbersome and often time-consuming process of setting those collaborations up. The data linkage part of those collaborations does not need to be time consuming. If it is well planned. Here we present two vastly different data linkage experiences and lessons learned, from our Scottish Safe Haven, supporting cross-agency social-housing and health projects. Collaborations between health board, local authorities, universities and charities, that set out to improve the health and well-being of vulnerable groups while simultaneously reducing the pressure on the local health system. Both projects required data linkage between the partners datasets; an NHS identified group of vulnerable people and a variety of property details and metrics, to identify those most in need. These datasets are inherently not homogenised nor set up for cross-sector data linkage. One project required a time-intensive multi-stage mixed-method linkage approach combining direct matching across multiple variable fields with fuzzy-matching techniques and manual verification. Where a year later, the other project was able to utilise a unique property identifier, that only had recently been introduced into the data of the national health care provider. Whilst we recognise, that a single unifying unique identifier is not always available for data linkage across different agencies. We propose a framework of recommendations to consider before the data sharing that can facilitate a smooth data linkage process.