Dataset reference · Google Open Buildings · building footprints
The source with the biggest gaps, stated plainly
Building footprints are the least complete source in the catalogue, and it is worth being direct about that rather than presenting a uniform table of sources. Footprints exist only where a publisher released them; whole regions legitimately return nothing. This page covers how the count is produced, why absence is reported as absence, and what a footprint count cannot tell you. Licence table: data sources. Computation: methodology.
How the footprint count is produced
The methodology page records the computation as a count only: LocateRight asks its stored spatial index how many footprint records fall inside the analysis radius, without fetching the individual geometries. Two footprint stores are queried and their counts summed:
Microsoft Global ML Building Footprints
One of two stores queried for the footprint count. Its contribution is reported under the
combined buildings source id with the dataset version attached.
Google Open Buildings
The second store. Published by Google Research using machine-learned extraction from satellite and aerial imagery.
Why summed, not deduplicated
The two stores cover largely different extents rather than duplicating the same records, so they are added rather than reconciled. Where both cover the same area, the result is a coverage-weighted count and not a de-duplicated inventory.
Why absence is reported as absence
Where neither store holds footprints covering the analysed point, the platform returns an
explicit NO_DATA_FOR_LOCATION state carrying the reason, rather than a number.
This distinction is the whole point of the block. A count of zero would assert that nothing is built. What the data actually says is that nothing is published there — a different claim, with a completely different implication for how you read an analysis. Reporting the first when you have only established the second would be a fabricated measurement, and it is the single most misleading thing a footprint-based product could do in a region it cannot see.
What a footprint count does not tell you
| Question | Answerable? | Why |
|---|---|---|
| How many buildings are in this radius? | Yes, where coverage exists | Directly counted, with the dataset version attached |
| Is this area densely built? | Yes, as built density | A count over a stated radius |
| Is this region covered at all? | Yes, explicitly | Reported as NO_DATA_FOR_LOCATION, never as zero |
| What is each building used for? | No | Footprint geometry carries no use, category or occupancy attribute |
| Is the retail here busy? | Not alone | Built density includes housing, warehousing and infrastructure |
| How old is a structure? | No | No construction-date attribute in the count |
| Is a specific building vacant? | No | A footprint is not an occupancy record |
| Is this the same building counted once? | Guaranteed no | The two stores are summed, not deduplicated |
Licence and provenance
Google Open Buildings is published under Creative Commons Attribution 4.0. LocateRight's licence registry records it as approved for commercial use, with attribution required and no share-alike obligation.
Google Open Buildings: FAQ
How does LocateRight count building footprints?
It returns the number of footprint records falling inside the analysis radius, counted directly from the stored spatial index rather than by fetching individual records, and summed across two footprint stores: Microsoft Global ML Building Footprints and Google Open Buildings. The two are added together rather than deduplicated, because they cover largely different extents rather than duplicating the same records. The dataset version is attached to the result.
Why is building coverage the least reliable source?
Because footprints only exist where a publisher has actually released them. Some regions legitimately return nothing at all. This is a gap in the underlying open data, not a failed search, and LocateRight reports it as an explicit NO_DATA_FOR_LOCATION state with its reason rather than as a count of zero.
What licence applies to Google Open Buildings?
Creative Commons Attribution 4.0. LocateRight's licence registry records it as approved for commercial use, with attribution required and no share-alike obligation.
Can I tell what a building is used for from this data?
No. The result is a count only, and footprint geometry carries no use, category or occupancy attribute. A footprint count tells you how much is built, not what it is used for. For what is actually on the ground, see the place records from Overture Places and OpenStreetMap.
Does a footprint count tell me if a location is commercially busy?
Not on its own. Footprint count measures built density, and built density includes housing, warehousing, sheds and infrastructure as well as retail. Read alongside place records it is informative; read alone it is easily over-interpreted, which is why it is presented as one block among several rather than as a verdict.
See the count, and see when it is absent
Run an analysis and inspect the footprint block with its dataset version and method attached — including the regions where the honest answer is "no data for this location".