Anyone who has tried to get their hands on geolocated address data in the UK will recognise the challenge. Navigating Ordnance Survey's catalogue is harder than it should be and even getting the right license can be difficult. At Rebalance Earth we use geolocated address data for identifying properties owned by pension funds and asset owners that are at risk of flooding so that we can propose nature-based flood management solutions to mitigate that risk.

As part of our flood risk reporting process, asset owners often send us a list of property addresses associated with individual commercial buildings, retail parks with several buildings, hospitals, or entire industrial parks. But for flood reporting we need polygons, and even the OS AddressBase data only gives us addresses as points. So, we had to find a better way of turning raw address strings into accurate, georeferenced polygons.

To do this, we built the Locator App: a CLI tool that stitches together several open-source geospatial datasets and uses a tiered address matching system to find or create a georeferenced polygon for every address in a list. The app works best for geolocating large areas with lots of buildings like retail parks, industrial estates, or commercial properties, rather than residential addresses.

This talk is about how we built the Locator App and why it matters beyond our own work. I want to share what tools we used, including Python, GDAL, pg_tgrm for fuzzy name matching, Claude for building the interface, and the open data we used such as OS Overture Places and OSM Land Use. I want attendees to leave inspired and thinking about how they can adopt an open-source approach to solving problems and break their own dependence on proprietary data.