Overture Maps Places
Point records for roughly 88 million businesses, landmarks, and other physical destinations worldwide.
About the data
Overture Maps Foundation builds the theme by conflating provider feeds: a gradient-boosted classifier scores candidate pairs on name, address, website, phone, category, and spatial distance, matched pairs are grouped into clusters, and the record from the source with the highest match count is promoted rather than attributes being merged across providers. Meta contributes roughly 60 million records and BrightQuery roughly 16 million, alongside Microsoft, Foursquare, PinMeTo, and smaller aggregators, with no OpenStreetMap data. Because one record is promoted instead of combined, the theme carries duplicates, a high junk rate, and low property completeness.
What's included
Global point coverage of schools, businesses, hospitals, religious organizations, landmarks, and mountain peaks, classified against a hierarchical taxonomy of roughly 2,300 categories plus a simpler basic category label drawn from about 280 terms. One row is a single place, carrying names, addresses, websites, phones, brand, an existence confidence score between 0 and 1, and a point geometry. Roughly 88 million rows in all, each keyed on a GERS ID that anchors the record across monthly releases.
Why we offer this
Most researchers reaching for this are not studying points of interest as such; they are measuring the retail, health care, or school environment around a household, a tract, or a study site and need a denominator for it. One taxonomy applied across every provider, at one grain, makes those counts comparable between countries rather than only within a single national business register.
Filter on confidence before counting. It scores existence only, is not calibrated to be comparable across providers, and raising the threshold trades coverage for precision.
More free datasets
Other open datasets hosted on Dewey