What is open data?

The definition: data anyone can access, use and share

The working definition comes from the Open Knowledge Foundation's Open Definition: open data is data that can be freely used, re-used and redistributed by anyone, subject at most to requirements of attribution and share-alike. Three conditions do the heavy lifting — the data must be available as a whole, in a machine-readable format, under a licence that permits reuse without discriminating against any user or any purpose.

That licence condition is the part people miss. Data you can look at on a website is not necessarily open data: if the terms forbid copying or commercial use, it is merely public-facing. Open data is defined by the rights you receive, not by whether a page happens to be visible.

• Available as a complete dataset, not drip-fed page by page
• Machine-readable, in open formats such as CSV, JSON or GeoJSON
• Licensed for any use, including commercial products
• Free of charge, or priced at no more than reasonable reproduction cost

The 5-star deployment scheme

Sir Tim Berners-Lee, inventor of the web, proposed a five-star scale for how openly data is published. One star simply means the data is on the web under an open licence, in any format — even a scanned PDF counts. Each further star makes the data easier for machines to process, reuse and connect to other data.

• ★ Available on the web under an open licence, in any format
• ★★ Machine-readable structured data — a spreadsheet instead of a scan
• ★★★ A non-proprietary format — CSV instead of Excel
• ★★★★ URIs identify each thing, so others can link to it (RDF)
• ★★★★★ Linked to other datasets, giving every record context

The main types of open data

Most open data falls into four families. Government data is the largest: budgets, spending, official statistics, transport timetables, legislation and business registers. Scientific data covers research outputs — climate measurements, genomic sequences, trial results — released by funders and institutions that increasingly mandate openness as a condition of funding.

Geographic data includes administrative boundaries, addresses, elevation and the community-built map of OpenStreetMap. The fourth family is commercial public data: facts businesses publish about themselves — company registrations, product catalogues, prices and storefront details — which anyone may read and, subject to law and the publisher's terms, reuse.

• Government: statistics, budgets, registers, transport
• Scientific: climate, health and research datasets
• Geographic: boundaries, addresses, maps
• Commercial public data: published business facts and storefronts

Major open data portals

Four portals anchor the ecosystem. data.gov is the United States federal catalogue; data.gov.uk covers UK public bodies; data.europa.eu aggregates European national portals into a single search; and World Bank Open Data publishes development indicators for economies worldwide. Around them sit thousands of national, regional and city portals publishing in the same spirit, which makes government open data the largest single body of reusable data in existence.

Licences: CC0, CC BY and the OGL

Three licences cover most open data. CC0 is a public-domain dedication: the publisher waives all rights, so the data can be used for anything, with no attribution required. CC BY is nearly as permissive but requires you to credit the source. The Open Government Licence (OGL), used across UK public data, works much like CC BY.

Some datasets add a share-alike condition — adaptations must be published under the same licence — and some government data carries specific terms of its own. The practical rule is simple: find the licence before you build on a dataset, and keep whatever attribution it asks for.

• CC0 — no rights reserved, no attribution required
• CC BY — reuse freely, credit the source
• OGL — UK government licence, attribution required
• ODbL — the OpenStreetMap licence: attribution plus share-alike

Open data vs open source

Open data and open source are cousins, not twins. Open source is about software: code published under licences such as MIT or GPL that let you run, study and modify programs. Open data is about datasets — the facts and figures that software consumes. A project can be both, neither, or one without the other: an open-source mapping tool may process closed data, and an open dataset may be explored with proprietary software. The two movements share a philosophy of reuse, but they govern different things and use different licences.

Open data vs commercial databases built on public data

If public facts are free, why do commercial databases built on them exist? Because raw public facts are rarely usable as published. They are scattered across millions of pages, uneven in quality, and stale the day after you collect them. The commercial product is the work done around the facts — collecting at scale, verifying each record, structuring it consistently, and keeping it current.

OpenData is one example. It fetches public storefronts and checks each one for ecommerce technology; as of the 15 Nov 2026 snapshot it holds 1,504,744 verified stores across 1,554,041 tracked domains, and every record carries its source URL and last-observed date. Only public business facts are kept, and only business emails stores publish for contact (info@, sales@, support@, or a named contact address) — never private addresses. Anyone can ask for a store to be removed; removals are honoured within 48 hours and permanently suppressed. Our comparison page sets out how this model differs from subscription databases.

How businesses actually use open data

Businesses use open and public data for unglamorous, valuable work. In every case the pattern is the same: the facts are public, and the value comes from having them complete, current and structured enough to act on.

• Market sizing — counting the stores, companies or outlets in a segment before entering it
• Lead generation — building prospect lists from public business facts instead of buying stale lists
• Competitive monitoring — tracking which platforms, apps and payment methods rivals adopt
• Risk and compliance — checking registrations and official registers before onboarding partners
• Enrichment — adding geography, demographics or technology data to existing customer records
• Location planning — using geographic and demographic data to choose where to open or deliver

Ecommerce: public storefront data as a commercial use case

Ecommerce is a clean example of the pattern. Every online store publishes facts about itself on its own storefront: the platform it runs on, the apps and payment gateways it uses, its currency and its contact details. Collected and verified, those facts become a store database — OpenData counts 269,688 verified Shopify stores alone, and its country lists range from 139,254 stores in the United States to 2,868 in Indonesia.

You can test the idea free. The store checker gives a free account 30 full-record unlocks a month, and there is one free sample file per industry. Records are priced per row — there are no subscriptions — and the index is refreshed monthly; a paid pull exports CSV plus a formatted Excel file.

Questions, answered straight

Next steps