Business Registry API Access by Country: The Global 2026 Map

If you've ever tried to build cross-border company verification in-house, you already know the punchline: every registry is different. Some hand you a clean REST API for free. Some make you pay €10 per company. Some have no API at all and rate-limit their web portal at 60 lookups an hour. A few don't even publish the data online.

This article maps the entire landscape — what's available, what it costs, and where the real friction sits. The table below is built from official registry documentation, our own production integrations, and recent firsthand work by KYB providers and registry-data researchers. Every row reflects the situation as of May 2026.

The honest global picture

The headline number that gets quoted in industry decks — "200+ countries have business registries" — is true, but it hides almost everything that matters. Having a registry is not the same as having an API. And having an API is not the same as having an API you'd want to put in production.

We grade every jurisdiction on a simple three-way scale. Either the registry runs an API or it doesn't — and where it does, we separate the ones anyone can sign up for from the ones that gate access behind credentials you may not be able to obtain.

A note on commercial aggregators. Most countries without an official API still have well-established private aggregators that operate paid APIs — Informa D&B in Spain, Cerved and CRIF in Italy, Bisnode and Creditreform in Germany, Altares in France, Experian and Equifax in the UK and US. These aren't government APIs; they're commercial intermediaries that license registry data, normalize it, and resell it. The categorization in the tables below scores official government APIs only. Where a major commercial aggregator is the practical access route, we name it in the comments column — that distinction matters when data provenance, audit trails, and update freshness are at stake.

A few things to internalize before you read the tables:

"API available" rarely means production-ready. Plenty of registries have an API in the technical sense — a JSON endpoint exists somewhere — but rate limits, authentication friction (national digital IDs, in-person KYC of the API user), or missing fields make it unusable at scale. Where we know about these traps, the comments column flags them.

Free is not the same as free-to-use commercially. The UK Companies House API is genuinely free with no commercial restrictions. France's Sirene is free and reusable. But several "free" registries restrict commercial reuse, require attribution, or limit downloads to non-automated traffic.

Most of the value is in coverage, not single-country depth. If you only need UK data, Companies House is excellent and you should use it directly. The moment you need three or more jurisdictions, the integration tax — different schemas, different IDs, different languages, different status taxonomies — becomes the largest line item in your project.

Open vs. closed: the regional view

Here is the same data, regrouped by the six conventional regions. For each region we show the share of jurisdictions with an open official API, a permission-only API, and no API at all. The full row count for each region is shown on the right.

Three findings jump out of this view.

Europe is the only region where "Yes" beats "No" by more than rounding. 38% of European jurisdictions run open APIs against 52% with none. Every other region is at least three-quarters closed. North America's 89% "No" rate is structural: the US has no federal register and 50 state systems with no APIs, dragging the regional average down despite Canada's federal exception.

Permission-only is concentrated in Europe and Africa. In Europe it reflects post-CJEU UBO gating and resident-entity requirements (Netherlands, Italy, Austria, Sweden). In Africa it reflects partner-accreditation regimes at the larger registries (Nigeria CAC, South Africa CIPC). North America, Latin America, and Asia barely use the model — they tend to skip the API entirely.

The Middle East and Africa look almost identical statistically at ~85% "No". But the underlying problem is different. Middle Eastern registries are well-funded but commercially gated (paid extracts, free-zone fragmentation). African registries are mostly underfunded and portal-bound. Same outcome, opposite root causes — and they require different commercial workarounds.

Throughput: not all "Yes" rows are equal

The categorical view masks a 100× operational range. Two registries can both be "Yes/Free" and differ by orders of magnitude in how many lookups you can run per hour. Here are the published rate limits for the most-integrated registries in production KYB stacks.

The 500× gap between France and Germany — neighbours in the EU, comparable economies, both with sophisticated regulatory regimes — is not a quirk. It's the consequence of a deliberate policy choice. France ran its statistical and registry data through API-first modernization in the 2010s. Germany kept paper-court infrastructure and never built a national replacement layer. If you're onboarding 50,000 entities a year, the UK ceiling is comfortable, the German one will force batching, queueing, and probably a commercial aggregator in the loop.

Data freshness: how stale is each registry, really?

A separate dimension that the "Yes/No" table doesn't capture: when changes happen at the company, how long until they're visible in the data? Real-time streams are rare. Most registries operate on filing cycles that introduce structural lag — director changes and address updates can drift 12–24 months between annual returns even in well-run jurisdictions.

The freshness chart sets up a point most KYB buyers miss: registry update cadence and data accuracy are different problems. Even the UK Companies House, which streams updates in real time, can show stale director data for a small company that hasn't filed in 11 months. Real-time means "real-time relative to the filing event" — not "real-time relative to ground truth". For ongoing monitoring this matters enormously: a counterparty can become insolvent, change controllers, or dissolve months before that information appears in any registry, anywhere.

This is the case for continuous monitoring on top of registry data, not annual re-screens. We unpack the workflow design in KYB vs. KYC in 2026.

Two patterns are worth calling out beyond the regional split.

Open-data Europe is still the global outlier. The UK, France, Norway, Denmark, Estonia, Latvia, Poland and the Czech Republic all run free, structured, commercially-usable APIs. That is a small club, and it doesn't include any of the world's three largest economies (US, China, Japan beyond the basic tax-ID lookup).

"No API" is not the same as "no data". Most jurisdictions without an official API still have a public web portal, paid extracts, or a commercial aggregator. The cost of using them at scale isn't licence fees — it's engineering. You're building and maintaining scrapers across changing HTML, CAPTCHAs, language barriers, and rate limits. The total cost of ownership of that approach is what drives most teams towards a single registry-data API in the first place.

What the data actually tells us

When we read the table end-to-end, eight patterns repeat. Most of them contradict the conventional wisdom about where company data is easiest to get.

1. The G7 has one open registry. Only one.

Across all seven of the world's largest advanced economies, only the United Kingdom offers a free, open, comprehensive business registry API. France comes close with INSEE Sirene, but Sirene is the statistical register — the legal-entity record sits in INPI's RNE, also free but operationally separate. Germany has no API. Italy is permission-only via reseller. Japan publishes a basic tax-ID lookup but the actual commercial register is paid and portal-bound. Canada's federal API exists but only covers federally-incorporated entities (about 10% of Canadian companies; the rest are provincial and have no API). And the United States has no federal register at all.

If you only consume the financial press, you'd think the UK was a regulatory laggard with messy filings and a Companies House that's been hauled over the coals by ECCTA. That's true on the quality dimension. On the access dimension, the UK runs circles around everyone else.

2. Eastern Europe quietly leads the open-data game

If we count free, anyone-can-sign-up registry APIs as the gold standard, 10 of the 28 globally come from former Eastern Bloc countries: Estonia, Latvia, Poland, Czech Republic, Slovakia, Slovenia, Bulgaria, Serbia, Ukraine, plus Russia (operationally complicated by sanctions). Add the Baltics' open UBO posture and you have a region that — counterintuitively — is more digitally open than France, Germany and Italy combined.

The reason is structural. Eastern European registries were largely digitized after 2000, after the late-90s wave of EU accession candidate reforms, and built on modern web stacks from the start. They didn't have to retire paper-court infrastructure the way Germany still does. Estonia's X-Road is the canonical example, but Poland's KRS, Czech ARES, and Slovenia's AJPES all follow the same logic.

3. The richer the country, the more likely the data is paid

Singapore (S$5.50 per profile), Hong Kong (per-search fees), Switzerland (cantonal paid extracts), Luxembourg (€10), Ireland (€2.50–€5), Netherlands (subscription + per-query), Sweden (paid APIs), and the entire Gulf (paid across the board) — wealthy jurisdictions consistently price registry data as a commercial product, not a public good. The opposite pattern shows up where you'd least expect it: Rwanda is free. Kazakhstan is free. Ukraine is free. Ecuador is free.

This isn't a values judgment — it's a budgeting one. Wealthy registries treat company-data licensing as a cost-recovery revenue line. Emerging-market registries that are still building digital infrastructure tend to publish openly because that's what the donor-funded modernization programs require.

4. The Gulf wealth bloc has zero open APIs

UAE, Saudi Arabia, Qatar, Kuwait, Bahrain, Oman — six economies, hundreds of billions in cross-border transactions, and not one open business registry API. Saudi Arabia comes closest with the Ministry of Commerce's National Business Center publishing structured data, but access is via authorized channels under contract. The UAE is the most extreme: each emirate's Department of Economic Development runs a separate licence register, and the country also has 45+ free zones (DIFC, ADGM, JAFZA, DMCC, RAKEZ, and dozens more), each operating its own register with its own access rules.

For compliance teams onboarding Gulf entities, this is the single largest hidden cost in any global KYB rollout. There's no shortcut. You either pay per-document via each emirate and free-zone authority, or you use a data provider that has aggregated them.

5. Africa is one country wide

Across roughly 48 sub-Saharan African jurisdictions, exactly one offers a free, open registry API: Rwanda's RDB. Nigeria's CAC and South Africa's CIPC offer permission-gated access. Every other African registry sits behind a web portal or paid extracts. The OHADA-region countries (17 West and Central African states sharing a unified commercial-law framework via the Common Court of Justice and Arbitration) all use the RCCM register format but expose data only through national portals — no shared API, no unified search.

6. The CJEU 2022 ruling created the "Permission only" category

Before November 2022, the Permission-only bucket was much smaller. The CJEU ruling that mandatory public access to UBO registers violated EU fundamental rights forced every EU member state to introduce legitimate-interest gating. Most never restored the open access that the 4th and 5th AMLDs had originally established. Several EU states (Austria, Italy, Netherlands among them) shifted further still — gating not just UBO data but core registry APIs behind obliged-entity status or in-country presence requirements.

The result, three and a half years on: AMLD6 obligations have stayed stringent, the data needed to satisfy them has become harder to access, and the burden of bridging that gap has shifted onto private data providers with negotiated access channels.

7. Latency and rate limits matter more than the "Yes/No" answer

Two registries can both be "Yes/Free" and still differ by an order of magnitude in production. The UK Companies House allows ~600 requests per 5-minute window per key. The German Handelsregister web portal — which is the closest thing Germany has — caps at 60 requests per hour. That's a 50× difference in throughput before you even start engineering. If you're onboarding 100,000 customers a year, the UK ceiling is generous and the German one will force batching, queueing, and probably commercial routing through a paid provider.

Even within the open tier: ARES (Czech) returns under 200ms typically. INSEE Sirene has variable latency depending on the endpoint and time of day. Companies House streams updates in real time. NZ Companies Office is fast but throttles aggressively on burst. Treat headline availability as necessary but not sufficient — instrument the upstream before committing to it.

8. The "Yes" rows are misleading without context

Look closer at any "Yes" row and you find caveats that change the production picture:

The UK has a free API — but ~60% of accounts at Companies House are still scanned PDFs as of 2026 and the API has no SLA. France has free APIs — but data is split across INSEE, INPI and Infogreffe with no single endpoint. Norway has a free API — but UBO data lives in a separate register with restricted access, and shareholders sit at the tax authority. Canada has a free federal API — but it covers only federally-incorporated entities; provincial coverage is portal-only. Japan has a free API — but it returns tax identity data, not the commercial register. Russia has a free API — but post-2022 sanctions make practical use complicated for Western firms.

There's no jurisdiction where the answer is "yes, here's everything, free, in one place, in English, with an SLA". The closest the world has come is Estonia, and Estonia has 1.3 million people.

The hidden costs people miss

API availability is one variable. Even where it exists, there are five others that usually decide whether a registry can sit inside a production KYB workflow.

1. Authentication friction

Many "open" registries require a national digital ID for full document access — SingPass for Singapore, BankID for the Nordics, NemID/MitID for Denmark, an EU/EEA equivalent for Norwegian documents. Without it, third-country institutions sit behind a permanent partial-access wall.

2. Language

The Handelsregister is in German. Registro Imprese is in Italian. NECIPS is in Chinese. Romanian, Hungarian, Czech, Greek, Bulgarian and most other Eastern European registries have only token English layers. Without machine translation and cross-language entity matching, structured data alone doesn't get you a comparable record.

3. Identifier instability

Germany has no national company number; the Handelsregister number changes when a company moves courts. The US has no federal company identifier; LLC IDs are state-specific. Brazil's CNPJ is federal but state-level Juntas Comerciais re-register entities for state purposes. Spain's NIF/CIF is stable, but the Registro Mercantil entry is provincial. Building a cross-jurisdiction graph means resolving identifiers — not just storing them.

4. Format quality

In the UK, ~60% of accounts at Companies House are still scanned PDFs. In Spain, annual accounts at the Mercantile Registry are PDF-only. In Germany, Bundesanzeiger filings vary widely. Structured XBRL filings are the exception, not the rule, outside the Nordics and a handful of leading EU jurisdictions.

5. UBO access after CJEU 2022

The November 2022 Court of Justice of the EU ruling shut down general public access to beneficial-ownership registers across the EU. Most member states now require a legitimate-interest demonstration. As of mid-2026, several states had still not transposed AMLD6, and the European Commission has opened infringement proceedings. UBO data is now harder, slower, and more lawyered-up to access than it was three years ago — even though the legal requirement to identify UBOs is stricter than ever. See our breakdown in KYB vs. KYC in 2026 for what that means for compliance design.

What this looks like at scale

If your business operates across two or three jurisdictions, you can integrate registries directly. Companies House is excellent. Sirene is excellent. ACRA is workable with the right credentials. You'll spend engineering effort, but it's manageable.

The moment you cross five or six jurisdictions — or you need any kind of UBO depth — the calculus changes. You are now maintaining:

Multiple authentication regimes. Multiple identifier systems. Multiple languages. Multiple status taxonomies (Germany alone has eight different company_status values, with free-text in-flight states layered on top). Multiple file formats. Multiple rate-limit regimes. Multiple update cadences. Multiple legal-basis regimes for the data you're holding.

And every one of those is a moving target. Companies House rolled out ECCTA identity verification through 2025 and 2026. Norway's UBO register went live in October 2024. The CJEU 2022 ruling reshaped half the EU overnight. The US Corporate Transparency Act is still tied up in litigation and policy reversals. You do not "build this once".