The turn to in-the-wild within HCI has given rise to an increasing concern around designing technologies which are available at large scale. Uniquely, at the intersection of public health and HCI, our work has supported the deployment of a mobile application, FeedFinder, over the last three years. We delineate the ground-work that was required to sustain this mobile application over the long-term. Focussing in particular on efforts made to engage institutions in taking ownership over FeedFinder and the data it provides, we reflect on the tensions that arose between users and civic institutions, particularly around 'what matters'. We provide a reflection on key requirements when designing a health data service and provide three lessons learnt which can guide researchers toward their own successful and productive long-term research deployments.