# OpenCorporates: coverage, licensing, and what you still have to build

OpenCorporates publishes primary-source registry data under an open licence, and publishes its own coverage gaps. Here is what that means if you are resolving messy real-world input.

Verified against primary sources on 2026-08-18. Source: https://savvyiq.ai/compare/opencorporates

**OpenCorporates:** Primary-source registry records from 140+ jurisdictions under an open licence, with attribute-level provenance back to the register and published per-jurisdiction coverage.

**SavvyIQ:** One resolved entity from a messy name plus free-text context, researched live when we do not already hold it, with per-field citations. Commercial use included.

## What OpenCorporates is

A public-benefit company that has published registry data from official sources under an open licence since 2010.

### Primary sources, by principle

Their published Data Principles commit to foundational data "sourced from official primary sources, not opaque third parties".

### The database is openly licensed

It is published under the Open Database License, which no commercial provider in this category offers.

### Provenance is the product

Each data item carries a provenance object with its source URL, source type and the date it was retrieved, and they are migrating to a "statement" model to attach that at the level of an individual fact.

### plei

In February 2026 they launched plei, a free open identifier minted from register records in the ISO 17442 format used by the Global LEI, currently in alpha and starting with the US.

### They publish their own coverage

The knowledge base gives record counts, per-field fill rates, refresh cadence and a last-updated date for every jurisdiction.

## What it costs, and what the licence permits

Two routes: free share-alike keys for public-benefit users, paid annual plans for everyone else.

### Three tiers, billed annually

About $3,040 a year for 500 calls a month, $8,930 for 2,500, and $16,230 for 5,000, with daily caps of 200, 500 and 1,000. Bulk delivery is quoted.

### That is about $0.51 a call, falling to $0.27

Plans are annual, so the monthly cap is a ceiling you pay for whether you use it or not.

### Free keys are share-alike keys

Their terms: keys are "free of charge to Permitted Users (share-alike keys) and for a fee under our non-share-alike API terms ... for example for use in commercial or proprietary applications".

### Companies are not Permitted Users

Verbatim: "financial institutions, corporations, government departments and regulatory authorities are not Permitted Users".

### Share-alike reaches your derived data

Combine their database with your own and "the resultant information must be published under the Attribution Share-Alike ODbL", unless you buy a non-share-alike licence.

### Free access is granted, not signed up for

Public-benefit access is by application, and whether you qualify "is in our sole discretion". There is no self-serve free commercial tier.

## Where OpenCorporates is the right answer

Several jobs where they excel.

### Verbatim from the government record

They publish the filing as the register holds it, linked back to the source it came from.

### Provenance and lineage

Their principles commit to "full end-to-end attribute-level audit trails", and the API delivers provenance objects to match.

### The open licence

A genuinely open database licence. Nobody else in this category offers one.

### Officer and control records

California alone carries 31.6 million officer records against 9.3 million companies.

### They publish their own gaps

Per-jurisdiction fill rates and refresh cadence, in public, including the unflattering numbers. No commercial vendor does this.

### An open identifier for the long tail

plei extends ISO 17442 identifiers to entities that will never register for an LEI, which they note covers "fewer than 3 million of the estimated 400+ million legal entities globally".

## What you still have to build

Every figure below is published by OpenCorporates.

### One record per registration, not one company

A business registered in forty states is forty records, and deciding which one is the company is left to you.

### They name the gap themselves

Their guide to sourcing registry data lists identity resolution as a workstream teams underestimate: "Without a single, nationwide ID, cross-state deduplication (same firm registered in multiple states, name variants, conversions/mergers) requires robust match logic and caution about false merges".

### Trade names are barely in the data

Alternative names, their term for a trading name or abbreviation, cover 0.36% of California records against 100% for legal names. A loose search still needs the string to be in the record.

### Delaware is the sharper case

The most-used US incorporation jurisdiction carries no alternative names and no company status at all, and a registered address on 38.77% of records.

### Industry codes are whatever the register files

US states do not file them, so they sit at 0.00%. The UK reaches 82.18% because Companies House does.

### Some jurisdictions are frozen

Spain shows 2.6 million records from the Spanish Mercantile Registry, last updated in August 2011, with new incorporations and company updates both marked offline.

### Their own principles ask for more

Principle five: "Collecting from original sources once a month, or even once a quarter, is no longer fit for purpose." California refreshes monthly, Delaware infrequently.

## Side by side

Two different products rather than two versions of one. Rows where OpenCorporates is stronger are marked.

| Capability | SavvyIQ | OpenCorporates |
| --- | --- | --- |
| What you can match on | Name plus free-text location and free-text context (v3 preview) | Name, optional jurisdiction code, optional date |
| What a match returns | One entity record, with its registrations linked | Ranked candidates against the registration records they hold |
| A company registered in many states | One profile | One record per registration |
| Messy input: trade name, DBA, partial string | Resolved from context when the registered name differs | Needs the string to be in the record. Alternative names are 0.36% in California |
| When you matched a branch, not the parent | Flagged on the record as a branch match | A separate record, scored down as a foreign branch |
| What comes back beyond identity | Description, website, headquarters, industry codes, ownership and legal form | The register record and its filings |
| Industry classification | NAICS and SIC with confidence | Whatever the register files. 0.00% in California, 82.18% in the UK |
| When the entity is not already held | Researched live at call time | No result. The database is the answer |
| Matching to a historical date | No | Best in class. Date-aware scoring promotes companies active on the date you give |
| Access to the raw filing | Not exposed today. We return the resolved entity and its registry identifiers | Best in class. The register record itself |
| Officer and control records | Officers where held | Best in class. 31.6M officer records in California alone |
| Provenance | Per-field basis with citations and an authority tier (v3 preview) | Attribute-level lineage back to the register |
| Explaining a match | Confidence 0-100, plus the factors behind it and the input that would sharpen it | A similarity score per candidate |
| Licensing | Commercial use included, no share-alike obligation | ODbL share-alike, or buy a non-share-alike licence |
| Free access | Self-serve signup | Best in class. Free at scale for journalists, NGOs and academics, by application |
| Commitment to start | None. Pay as you go, no minimum | Annual plan, or a quote for bulk delivery |

OpenCorporates capabilities are drawn from its published API reference, reconciliation documentation, terms of use, pricing page and per-jurisdiction coverage tables, all retrieved on 18 August 2026. Coverage percentages are OpenCorporates figures for its own data. Prices are published in pounds sterling and converted at the European Central Bank reference rate for 18 August 2026, £1 to $1.3526. No precision or recall benchmark is published for either matching service, and we have not run a head-to-head test between them.

## How matching works

Two endpoints, two different jobs.

**The same problem, put to both APIs: which company called Delta?**

- OpenCorporates reconcile: `query=Delta
jurisdiction_code=us_de
date=2026-08-18`
- SavvyIQ: `name=Delta
location=Atlanta, GA
context=airline`

Their call has no field for what a company does. Delta is also a faucet maker and a dental plan. Both request shapes are each side’s own published parameters.

### Three things to match on

A name, an optional jurisdiction code, an optional date. No address, domain, identifier or free-text context.

### Search is deliberately loose

Their docs: "the search is deliberately quite loose ... allowing other words to be present too", so "Bank Barclays" finds "Barclays Bank".

### Matching is a reconciliation endpoint

Built for OpenRefine, the desktop data-cleaning tool. It returns ranked candidates for a person or an application to choose between.

### The score is name similarity

Scored on similarity to the name you sent, then adjusted down for inactive companies and foreign branches.

### Jurisdiction is how you disambiguate

Optional, and their guidance is that restricting to one is "a good idea" because names "are quite common across different locations". Supplying it is on you.

### Date-aware matching is a genuine strength

A date promotes companies active at the time, which is the right tool for historical datasets. We do not offer it.

## Which one to choose

### Choose OpenCorporates when

- You want the register record itself, filing by filing.
- Officer, control and ownership records are the core of the work.
- You need an auditable trail from every field back to the register.
- You are matching old records to the company that was active at the time.
- You qualify for their public-benefit access as a journalist, NGO or academic.

### Choose SavvyIQ when

- Your input is messy: a trade name, a DBA, a partial string or a misspelling.
- The companies you care about are in the long tail, not the household names.
- You do not know which jurisdiction the company is registered in.
- One company registered in many states needs to come back as one profile.
- You need more than the register holds: a description, a website, industry codes, ownership and legal form.
- Your licence has to permit commercial and proprietary use without share-alike.

## Common questions

### Is the OpenCorporates API free?

Only for Permitted Users, who get share-alike keys. Commercial and proprietary use is chargeable under their non-share-alike terms, and OpenCorporates states that financial institutions, corporations, government departments and regulatory authorities are not Permitted Users. Public-benefit access is granted by application at their sole discretion, not by signing up.

### Can I use OpenCorporates data in a commercial product?

Yes, with a paid non-share-alike licence. The database is under the Open Database License, and their terms state that if you combine the information with your own data, the resultant information must be published under the Attribution Share-Alike ODbL. If share-alike does not work for your product, the non-share-alike licence is the route.

### What does the OpenCorporates API cost?

Three published annual tiers: about $3,040 a year for 500 API calls a month, $8,930 for 2,500, and $16,230 for 5,000, with daily caps of 200, 500 and 1,000. That is about $0.51 a call at the entry tier falling to $0.27 at the top one. Bulk delivery is quoted. Prices are published in pounds at £2,250, £6,600 and £12,000.

### Does OpenCorporates do entity resolution?

It offers a reconciliation endpoint built for OpenRefine, the desktop data-cleaning tool. The call accepts a company name, an optional jurisdiction code and an optional date, and returns ranked candidates scored on name similarity, adjusted down for inactive companies and foreign branches. There is no field for an address, a domain, an identifier or free-text context, so choosing between same-name candidates is left to you.

### Is OpenCorporates search exact match?

No. Their documentation says the search is "deliberately quite loose, requiring the returned companies to have all the searched-for words (in any order), but allowing other words to be present too". The limit is not the tokeniser, it is the record: a loose search still needs the string to exist in the data, and alternative names such as trading names cover 0.36% of California records.

### What is plei?

plei, the proto Legal Entity Identifier, is a free open identifier OpenCorporates launched in February 2026. It uses the ISO 17442 format shared with the Global LEI and is minted from official register records rather than requested by the entity, which is how it reaches businesses that will never register for an LEI. It is in alpha, launching with US entities.

### How fresh is OpenCorporates data?

It varies by jurisdiction, and they publish the cadence for each one. Bands run from daily down to "Offline (no updates in over 180 days)", with an "Archive" state for registers they no longer have access to. California expects new incorporations weekly and updates monthly. Delaware is listed as infrequent.

### How accurate is OpenCorporates matching?

No precision or recall benchmark is published for it, and we have not run a head-to-head test against it. What is published is checkable: the reconcile call matches on a name with an optional jurisdiction and date, and OpenCorporates publishes per-jurisdiction fill rates showing which fields exist to match against.

## Sources

- OpenCorporates, Terms of use, including the database licence and API key terms (https://opencorporates.com/terms-of-use-2/), retrieved 2026-08-18
- OpenCorporates, API and bulk pricing (published in GBP) (https://opencorporates.com/pricing/), retrieved 2026-08-18
- OpenCorporates API Reference, version 0.4.8 (https://api.opencorporates.com/documentation/API-Reference), retrieved 2026-08-18
- OpenCorporates, Open Refine Reconciliation API documentation, version 0.4.8 (https://api.opencorporates.com/documentation/Open-Refine-Reconciliation-API), retrieved 2026-08-18
- OpenCorporates knowledge base, USA: California coverage (https://knowledge.opencorporates.com/knowledge-base/us_ca/), retrieved 2026-08-18
- OpenCorporates knowledge base, USA: Delaware coverage (https://knowledge.opencorporates.com/knowledge-base/us_de/), retrieved 2026-08-18
- OpenCorporates knowledge base, United Kingdom coverage (https://knowledge.opencorporates.com/knowledge-base/gb/), retrieved 2026-08-18
- OpenCorporates knowledge base, data coverage overview and jurisdiction status legend (https://knowledge.opencorporates.com/knowledge-base/overview/), retrieved 2026-08-18
- OpenCorporates, Legal-Entity Data Principles (https://opencorporates.com/legal-entity-data-principles/), retrieved 2026-08-18
- plei, the proto Legal Entity Identifier, launched by OpenCorporates (https://p-lei.com/), retrieved 2026-08-18
- OpenCorporates blog, What you need to know before sourcing data directly from US state registries (15 September 2025) (https://blog.opencorporates.com/2025/09/15/sourcing-data-directly-from-us-state-registries/), retrieved 2026-08-18
- OpenCorporates knowledge base, Spain coverage (https://knowledge.opencorporates.com/knowledge-base/es/), retrieved 2026-08-18

OpenCorporates is a trademark of its owner. This page describes OpenCorporates from public sources only and is not endorsed by or affiliated with them.
