Data sources · licences · what each one cannot tell you
Every source, every licence, and where each one stops
A dataset without a stated licence is a dataset you cannot safely build on. This page lists what LocateRight ingests, who publishes it, under what terms, whether those terms have been reviewed, and — for each source — the specific thing it cannot tell you. The methodology page covers how the measurements themselves are produced.
The rule we apply to every licence
Open, free and available are not the same as approved for commercial use. A licence only becomes commercially usable here once its terms have actually been read and recorded as APPROVED. Anything unclear stays REVIEW_REQUIRED and is held out of the free-commercial core entirely. The cost of that rule is that some perfectly usable data stays unused. We accept that cost: an unreviewed licence treated as permissive is how a product ends up redistributing something it was never entitled to.
Terms read and recorded. Commercial use permitted, subject to the source's own attribution and share-alike conditions.
Not yet cleared. Excluded from the free-commercial core until someone finishes reading the terms.
Source, dataset version and retrieval timestamp travel with every measurement and are attached to exports.
Sources that power the analysis
These are the only sources that produce the measurements you see when you analyse a location. Links go to the publisher's own pages, so the terms can be read at source.
| Source | Publisher | What it contributes | Licence | Status |
|---|---|---|---|---|
| Overture Places | Overture Maps | Business and place records behind the business-density and places view. | CDLA-Permissive-2.0 | Approved |
| OpenStreetMap POI | OpenStreetMap Foundation | Points of interest, contributing to the same place measurements. | ODbL 1.0 | Approved |
| WorldPop | WorldPop | The population figures in the population block. | CC BY 4.0 | Approved |
| ESA WorldCover | ESA | The land-cover block. | CC BY 4.0 | Approved |
| Open Buildings | Google Research | Building-footprint counts, only where footprints have actually been published for the area. | CC BY 4.0 | Approved |
| Natural Earth | Natural Earth | The country boundary drawn on the map. | CC0 1.0 | Approved |
| GeoNames | GeoNames | The nearest-city context line only. | CC BY 4.0 | Approved |
| Google Places | Google Maps Platform | Queried live at request time when enabled. Never stored, per Google's terms of service. | Google Maps Platform terms — live use only | Approved |
Nightlights and weather datasets exist in the backend but the analysis view does not request them.
Catalogue datasets, and what they are not
The catalogue, reachable through the API documentation, lists further datasets with their records, extents and provenance. Stating this precisely matters more than sounding comprehensive: these are catalogue metadata and they do not drive the numbers above. A dataset appearing in the catalogue tells you it is described and reachable through the API. It does not tell you it is wired into an analysis.
| Dataset group | Publisher | Licence | Status |
|---|---|---|---|
| Global Human Settlement Layer (GHSL) | Copernicus / EU JRC | CC BY 4.0 | Approved |
| VIIRS / DMSP night lights | NOAA NGDC | Public domain (US federal work) | Approved |
| NOAA / NCEI climate and weather | NOAA NCEI | Public domain (US federal work) | Approved |
| Sentinel earth observation | ESA Copernicus | Copernicus data licence | Approved |
| OurAirports | OurAirports | Public domain | Approved |
| Wikidata | Wikimedia Foundation | CC0 1.0 | Approved |
| ERA5 reanalysis | Copernicus Climate Data Store | Copernicus Climate Data Store terms — reviewed by legal before commercial redistribution | Review required |
| GTFS transit feeds | Per-feed publisher | Each feed carries its own terms and must be recorded per feed. Do not assume. | Review required |
What each source cannot tell you
This is the part most data pages leave out. Each source has a blind spot, and knowing it in advance is more useful than discovering it later.
Population grids are estimates, not censuses
WorldPop produces modelled gridded population, disaggregated from census and survey inputs. It is a strong national-scale estimate and a poor substitute for a street-level count. It tells you how much people live around a point; it does not tell you who they are, when they are there, or whether they are your customers.
Place records measure mapping, not existence
How complete a place layer is tracks how thoroughly a region has been mapped, not how many businesses operate there. A dense city and a sparse rural area differ in record count partly because they are surveyed differently. Absence from a layer is weak evidence of absence in reality.
Building footprints are published unevenly
Footprints exist only where a publisher has released them. Some regions return no data at all, and that is a gap in the upstream open data rather than a failed search. LocateRight marks this as coverage-dependent rather than reporting a zero as if it were a measurement.
Land cover is not land use
ESA WorldCover classifies surface cover — vegetation, water, built-up, bare ground. It is a 10 m classification with its own year and error characteristics. It does not know what a building is used for, whether a plot is let, or whether a site is commercially viable.
Opening hours are often unpublished
Where an hours field exists it is shown as published. It is not verified against the business, not confirmed current, and absent for most records. Treat a published hours value as a claim by the source, not as a verified fact.
No source here measures demand
None of these datasets observes spending, footfall or intent. That is deliberate: the platform draws a line at measurement and stops. Read the methodology for exactly which derived figures exist and which questions they cannot answer.
How provenance travels with the data
Every measurement is stored with three fields attached, and they are not cosmetic. They are what makes a figure checkable by someone who was not present when it was produced.
A worked example from a live record, not a mock-up. The same three fields are attached to exported data, so a figure keeps its origin after it leaves the platform.
Because the version is recorded, two analyses of the same point taken on different dates can be compared honestly: if a number moved, either the world changed or the upstream dataset did, and the version stamp tells you which. Without that field, a change in a published figure is indistinguishable from a change in the measurement.
Using this data yourself
Two routes, with different obligations.
- Reading it here. Every figure on the platform is displayed with its source, version and retrieval time. Reading is unrestricted.
- Exporting or redistributing it. Obligations travel with the data. CC BY 4.0 sources require attribution; ODbL additionally imposes share-alike on derivatives. Google Places data is queried live under Google Maps Platform terms and is never stored or redistributed. The licence column above is the one to read before you publish anything built on an export.
If you need a guarantee for a commercial decision, the per-record licence attached to your export is the authoritative statement — it reflects the actual snapshot you received rather than this page's summary of it.
Data sources and licences: FAQ
Can I use LocateRight data commercially?
It depends on the source, and the answer is per source rather than per product. Each dataset below carries a licence and a review status. A licence is only treated as free for commercial use once it has been reviewed and marked APPROVED; anything unclear stays REVIEW_REQUIRED and is kept out of the free-commercial core. Attribution and share-alike obligations differ per source, and they travel with the data into exports. Check the specific source before you redistribute anything.
Why does a licence say REVIEW_REQUIRED?
Because nobody has finished checking it, not because it is permissive. A licence is only treated as free for commercial use when its terms have been read and recorded as APPROVED. ERA5 and per-feed GTFS data are currently in this state, so they are excluded from the free-commercial core. Treating an unreviewed licence as permissive is exactly the mistake this status exists to prevent.
What is the difference between a dataset in the catalog and one that powers the analysis?
A catalog dataset is described metadata: it has records, extents and provenance entries, but it does not drive the numbers you see when you analyse a location. Only the sources listed under What powers the analysis produce those measurements. The catalog holds more, and the distinction is stated rather than blurred.
How fresh is each dataset?
Each source updates on its own publisher's release cadence, and LocateRight does not control that schedule. What LocateRight does control is the record of it: every measurement carries the dataset version it came from and the timestamp it was retrieved, so you can tell which release you are reading and when it was fetched. Freshness is a property of the upstream publisher, not a claim this platform makes on its behalf.
Which source is least complete, and why does that matter?
Building footprints are the most coverage-dependent source: they exist only where a publisher has actually released them, so some regions legitimately return nothing. Place records inherit the same problem in a milder form, because completeness tracks how thoroughly a region has been mapped rather than how many businesses exist there. A gap in these sources is a gap in the underlying open data, not a gap in the search you ran.
Does LocateRight show which source a number came from?
Yes. Every measurement and record carries its source, its dataset version and its retrieval timestamp, and exports attach the same information. This is the platform's core commitment: you should be able to trace any figure back to the release it came from rather than taking it on trust.
Read each source in detail
The table above gives you the licence position in one line. Each reference page below takes one dataset further: what it actually measures, at what resolution, how LocateRight uses it, what it cannot tell you, and the licence terms in full. They are written to be cited — each claim traces to the licence registry or to the code path that consumes the data.
- WorldPop — gridded population at roughly 100 m: the radius sum, the resolution limit, and why a headcount is not a customer count.
- Overture Places — place records, how duplicates with OpenStreetMap are reconciled into one set, and CDLA Permissive 2.0.
- OpenStreetMap — and the ODbL share-alike obligation, the one source whose licence travels with your export.
- ESA WorldCover — 10 m land cover classes, annual releases, and why land cover is not land use.
- Google Open Buildings — building footprints, and why this is the source with the biggest coverage gaps in the catalogue.
For how these are computed inside an analysis, see the methodology page and, for the accessibility block specifically, the drive-time methodology.
Read the method, then look at a place
The methodology page sets out how each measurement is produced and where each stage can fail. Once you know what the figures mean and what they cannot mean, put a real candidate location on the map and check the provenance yourself.
Found a licence error, a coverage gap or a figure that does not match its own provenance line? Report it on the contact page — corrections travel with the analysis reference, and the export carries the source, dataset version and retrieval timestamp needed to check it.
Working with an AI agent that asks for a machine-readable index? /llms.txt lists every public page on this site with a one-line summary of what each one covers. It is a convenience file, not a Google ranking factor.