Company Profile from Wikidata + GLEIF (LEI)

Give it company names, LEIs or Wikidata IDs and get one structured profile per company, built from two official open sources: the GLEIF Legal Entity Identifier registry (legal name, status and renewal of the LEI, legal form, register number, addresses, parent companies) and Wikidata (website, industry, size, revenue, stock listings, CEO, social links). No API key, no scraping of company websites, no login.

For KYC and compliance checks, sales intelligence, CRM enrichment and research. What makes it safe to use in a pipeline: the two sources are joined only through the LEI written on the Wikidata item; a name that fits several companies comes back as ambiguous with the candidates, never as a guess; every record says how it was matched (matchConfidence + matchExplanation) and carries the exact source URLs.

• Pay per profile found: $0.0012 ($1.20 per 1,000). Not found, ambiguous, invalid input, repeats and runs where a source did not answer are free.
• Official open data. GLEIF data is CC0; Wikidata's structured data is CC0. Polite by design: one request at a time per source, with a pause, Retry-After honoured.
• Not affiliated with, or endorsed by, GLEIF or the Wikimedia Foundation. The GLEIF logo is not used.

Input

Field	Default	What it does
companies	Apple Inc. / LEI / Q312	One entry per company, up to 1,000: a name (Apple Inc.), a LEI (HWUPKR0MPOU8FGXBT394, any case) or a Wikidata ID (Q312 or its wikidata.org URL). The kind is recognised by itself. A LEI whose check digits (ISO 17442) are wrong, or with 19 or 21 characters, is reported as invalid_input and never searched as a name
country	none	Two-letter ISO code (US, DE, IT). For names only: keeps the entities of that country, which settles most ambiguous names
includeRelationships	on	Add the direct and ultimate parent from GLEIF. Costs two more GLEIF requests per company (a slower run, same price)
maxCandidates	3	How many look-alike entities to list in candidates for a name search (0-10)

Repeats (same LEI, same Wikidata ID, same name written another way) are collapsed. If you ask for the same company in different words (a name and its LEI), the second profile comes back as duplicate: delivered, not billed.

What you get

One record per entry, same keys every time (null or an empty list when the sources say nothing; nothing is invented).

Field	Meaning
query, matchType	Your entry and how it was read: lei, qid or name
status, statusMessage	found (billed) · duplicate · partial (a source did not answer: delivered, free) · ambiguous · not_found · invalid_input · error. The message says why in plain words
matchConfidence, matchExplanation	high = you gave an identifier, or an exact name that no other entity shares. medium = chosen among several for a reason stated in the explanation, or matched on Wikidata only, or joined to Wikidata by name
lei, legalName, otherNames, name	The LEI, the legal name in GLEIF, other names (previous names, translations), and a display name (Latin script when the legal name is not)
entityStatus, entityCategory, registrationStatus, leiLapsed, renewalOverdue, nextRenewalDate, initialRegistrationDate, lastUpdateDate	ACTIVE/INACTIVE company; entityCategory = GENERAL (a legal entity), BRANCH (an office of another entity: never chosen over the legal entity of the same name), FUND...; LEI registration (ISSUED, LAPSED, RETIRED...); leiLapsed = GLEIF marked it lapsed; renewalOverdue = still ISSUED but the renewal date has passed
legalForm, jurisdiction, registrationAuthority	ISO 20275 code and English name; jurisdiction; the business register and the company's number in it
legalAddress, headquartersAddress	Lines, city, region, country, postal code, one-line formatted
directParent, ultimateParent, successorLei	{lei, name, exception}: the parent, or the reason GLEIF states for having none (NATURAL_PERSONS, NO_KNOWN_PERSON, NON_CONSOLIDATING, NON_PUBLIC...). null when includeRelationships is off
wikidataId, wikidataLinkedBy	The item and how it was tied to the record: lei (same LEI, property P1278), qid (you gave it), name (name + country only: check it)
description, website, industry, inception	From Wikidata. Dates keep Wikidata's precision (1976-04-01, 1976-04 or 1976)
employees, revenue	The most recent value with its date: {value, asOf} and {amount, currency, asOf}. Wikidata's figures can be years old: check asOf
stockExchanges, isin, ceo	Current listings with ticker, ISIN of the shares, current CEO (statements with an end date are left out)
country, countryCode, logoUrl, social, wikipediaUrl	Country (Wikidata); a link to the logo file on Wikimedia Commons; LinkedIn, X, Facebook, Instagram, YouTube, GitHub, TikTok links built from Wikidata identifiers; English Wikipedia article
candidates	Look-alike entities (lei, wikidataId, name, country, entityStatus, entityCategory, matchScore): the choices for an ambiguous name
warnings, sources, gleifDataDate, charged, checkedAt	Things to check; the exact GLEIF and Wikidata URLs; the date of the GLEIF data the record comes from; whether it was billed; when it was built

Example: a found profile (real run, input HWUPKR0MPOU8FGXBT394)

{
  "query": "HWUPKR0MPOU8FGXBT394",
  "matchType": "lei",
  "status": "found",
  "statusMessage": "Profile found.",
  "matchConfidence": "high",
  "matchExplanation": "The LEI was given and exists in GLEIF. Wikidata item Q312 carries the same LEI (P1278).",
  "name": "Apple Inc.",
  "lei": "HWUPKR0MPOU8FGXBT394",
  "legalName": "Apple Inc.",
  "otherNames": [
    {
      "name": "Apple Computer, Inc.",
      "type": "PREVIOUS_LEGAL_NAME",
      "language": "en"
    }
  ],
  "entityStatus": "ACTIVE",
  "entityCategory": "GENERAL",
  "registrationStatus": "ISSUED",
  "leiLapsed": false,
  "renewalOverdue": false,
  "nextRenewalDate": "2027-03-08T17:27:20Z",
  "initialRegistrationDate": "2012-06-06T15:53:00Z",
  "lastUpdateDate": "2026-03-03T16:34:33Z",
  "legalForm": {
    "code": "H1UM",
    "name": "For-Profit Corporation General Stock"
  },
  "jurisdiction": "US-CA",
  "registrationAuthority": {
    "id": "RA000598",
    "name": "Business Entity Records",
    "website": "https://businesssearch.sos.ca.gov/",
    "entityId": "806592"
  },
  "legalAddress": {
    "lines": [
      "C/O C T Corporation System",
      "330 N. Brand Blvd",
      "Suite 700"
    ],
    "city": "Glendale",
    "region": "US-CA",
    "country": "US",
    "postalCode": "91203",
    "formatted": "C/O C T Corporation System, 330 N. Brand Blvd, Suite 700, 91203 Glendale, US-CA, US"
  },
  "headquartersAddress": {
    "lines": [
      "One Apple Park Way"
    ],
    "city": "Cupertino",
    "region": "US-CA",
    "country": "US",
    "postalCode": "95014",
    "formatted": "One Apple Park Way, 95014 Cupertino, US-CA, US"
  },
  "directParent": {
    "lei": null,
    "name": null,
    "exception": "NATURAL_PERSONS"
  },
  "ultimateParent": {
    "lei": null,
    "name": null,
    "exception": "NATURAL_PERSONS"
  },
  "successorLei": null,
  "wikidataId": "Q312",
  "wikidataLinkedBy": "lei",
  "description": "American multinational technology company based in Cupertino, California",
  "website": "https://apple.com/",
  "industry": [
    "software industry",
    "consumer electronics industry",
    "digital distribution"
  ],
  "inception": "1976-04-01",
  "employees": {
    "value": 164000,
    "asOf": "2022"
  },
  "revenue": {
    "amount": 416161000000,
    "currency": "USD",
    "asOf": "2025"
  },
  "stockExchanges": [
    {
      "exchange": "Nasdaq",
      "ticker": "AAPL"
    }
  ],
  "isin": [
    "US0378331005"
  ],
  "ceo": "John Ternus",
  "country": "United States",
  "countryCode": "US",
  "logoUrl": "https://commons.wikimedia.org/wiki/Special:FilePath/Apple_logo_black.svg",
  "social": {
    "linkedin": "https://www.linkedin.com/company/apple",
    "x": "https://x.com/Apple",
    "facebook": "https://www.facebook.com/apple",
    "instagram": "https://www.instagram.com/apple/",
    "youtube": "https://www.youtube.com/channel/UCE_M8A5yxnLfW0KghEeajjw",
    "github": "https://github.com/apple",
    "tiktok": "https://www.tiktok.com/@apple"
  },
  "wikipediaUrl": "https://en.wikipedia.org/wiki/Apple_Inc.",
  "candidates": [],
  "warnings": [],
  "sources": {
    "gleif": "https://api.gleif.org/api/v1/lei-records/HWUPKR0MPOU8FGXBT394",
    "wikidata": "https://www.wikidata.org/wiki/Q312"
  },
  "gleifDataDate": "2026-10-19T00:00:00Z",
  "charged": true,
  "checkedAt": "2026-10-19T09:21:49.223Z"
}

(otherNames and industry are shortened here; the real record lists every entry.)

Example: an ambiguous name (real run, input Meta): free, with the choices

{
  "query": "Meta",
  "matchType": "name",
  "status": "ambiguous",
  "statusMessage": "19 different entities carry this name. Give a country (ISO code) or run again with the LEI or Wikidata ID of the one you mean.",
  "matchConfidence": null,
  "lei": null,
  "wikidataId": null,
  "candidates": [
    {
      "lei": "984500Y657CE399B1547",
      "wikidataId": null,
      "name": "META",
      "country": "DK",
      "entityStatus": "ACTIVE",
      "matchScore": 1
    },
    {
      "lei": null,
      "wikidataId": "Q15813961",
      "name": "Meta",
      "country": null,
      "entityStatus": null,
      "matchScore": 1
    }
  ],
  "charged": false
}

Run again with the LEI or Wikidata ID of the one you mean (or add country).

Example: a mistyped LEI (real run, input 254900HLPLIRDQPWL69): free

{
  "query": "254900HLPLIRDQPWL69",
  "status": "invalid_input",
  "statusMessage": "\"254900HLPLIRDQPWL69\" has 19 characters and looks like a LEI, but a LEI has exactly 20: check for a missing or extra character",
  "charged": false
}

What you pay, and what you don't

$0.0012 per profile with status: found (event company): $1.20 per 1,000.

Run	Billed	Cost
7 Italian companies by name, all found (real run)	7	$0.0084
100 LEIs, all found	100	$0.12
1,000 companies, 850 found, 150 unclear or unknown	850	$1.02

Free, always: not_found, ambiguous (with candidates), invalid_input, duplicate, partial and error (a source did not answer: run it again later), the SUMMARY record.

Your spending limit is exact. With a maximum cost per run, the Actor checks the budget before every billed record and stops when it is spent (SUMMARY.stopReason: spending limit reached). It never charges past it.

How long a run takes

The two sources are read politely: GLEIF one request every 1.25 s (the GLEIF API publishes no limit; this stays under 50 per minute), Wikidata one at a time. A LEI takes about 5 s (several LEIs are read from GLEIF in blocks of 50), a name about 5 to 8 s with parents. Plan about 10 minutes per 100 companies; includeRelationships: false is faster. 256 MB of memory is enough.

FAQ

Where does each field come from? Everything about the legal entity (lei to successorLei) comes from GLEIF; everything about the business (website to social) comes from Wikidata. sources has the exact URLs. The GLEIF data is the "Golden Copy" published daily (gleifDataDate); Wikidata is edited by volunteers and can be out of date or empty for smaller companies.

How are a name and a LEI tied together? Through property P1278 ("LEI") on the Wikidata item. If the item carries the LEI, wikidataLinkedBy is lei. If a company exists in both sources but the Wikidata item has no LEI, the Actor joins them by name and country, only when the match is one to one, and says so (wikidataLinkedBy: "name", a warning, medium). It never joins by similarity alone.

Why ambiguous instead of the best guess? In a compliance or CRM flow a wrong company is worse than no answer. When several entities fit (the same name in other countries, a group holding and its operating company, funds), you get the candidates and decide. A different legal form counts as a different entity ("Ferrari N.V." is not "Ferrari S.p.A."). An entity whose registered legal name is exactly your input beats one that only has your input as an alias ("Roche Holding AG" is Roche Holding AG, not F. Hoffmann-La Roche AG); when two entities have that exact name ("UBS AG" in Switzerland and in the US) the answer is ambiguous until you give a country. A branch (entityCategory BRANCH: Intesa Sanpaolo's Paris office has the exact name of the bank) is never chosen over the legal entity with that name; it stays in candidates, marked. Legal forms are compared in every spelling ("S.A.B. de C.V." = "S A B DE C V" = "SAB de CV", "S/A", "Ltda."). A LEI that Wikidata gives to an item is trusted unless it is lapsed or retired AND registered in another country than the item (Wikidata has such mistakes: the Mexican Cemex carried the lapsed LEI of CEMEX, Inc., US); then the valid entity of the item's country is chosen, with a warning, or the answer is ambiguous. GLEIF records marked DUPLICATE are ignored. The name search also tries the legal form written without dots ("S.p.A." and "SPA" are different words for the GLEIF search).

What does it NOT do? No people, no contacts, no emails or phone numbers, no financial statements (only the revenue and headcount that Wikidata holds, with their dates), no ownership beyond the direct and ultimate parent, no sanctions or PEP screening, no scraping of company websites. It is not a replacement for an official extract of a business register.

Is the LEI status legally binding? No. The Actor reports what GLEIF publishes; GLEIF provides its data "as is". For a decision with legal effect, check the record at GLEIF.

Licences. GLEIF: "The data available through the Access Service are provided under the CC0 licence" (https://www.gleif.org/en/meta/lei-data-terms-of-use). Wikidata structured data: CC0 (https://www.wikidata.org/wiki/Wikidata:Licensing). Logos and images are not downloaded: logoUrl is only a link to the file page on Wikimedia Commons, where each file has its own licence: check it before any reuse. The Actor identifies itself as CompanyProfileBot/1.0 with a contact, as the Wikimedia User-Agent policy asks.

Which companies are covered? GLEIF holds about 3.4 million entities that asked for a LEI (all large and listed companies, banks, funds, many others); Wikidata covers notable companies. A small business with neither is not_found.

Limits, stated plainly

• A name is matched on its words: the register must contain them. Abbreviations ("IBM" for "International Business Machines") work only when Wikidata knows the abbreviation as a name of the company.
• Up to 300 GLEIF records are examined per name (three pages); a very common name may leave the right entity out: add country or use the LEI.
• The Wikidata part reads English labels. Revenue and headcount are the latest value on Wikidata, often not for the current year.
• Ownership is GLEIF's "Level 2" data: many companies report an exception instead of a parent.

Integrate

Through the Apify API:

GET https://api.apify.com/v2/datasets/<datasetId>/items?format=json&clean=true

JavaScript:

import { ApifyClient } from 'apify-client';
const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('far003/company-profile-lei-wikidata').call({ companies: ['Siemens AG', '5493006W3QUS5LMH6R84', 'Q53268'] });
const { items } = await client.dataset(run.defaultDatasetId).listItems();
for (const c of items) console.log(c.query, c.status, c.lei, c.entityStatus, c.registrationStatus);

Python:

from apify_client import ApifyClient
client = ApifyClient("<APIFY_TOKEN>")
run = client.actor("far003/company-profile-lei-wikidata").call(run_input={"companies": ["Siemens AG", "Q53268"], "country": "DE"})
for c in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(c["query"], c["status"], c["lei"], c["matchConfidence"])

To keep only what needs a human look, filter on status (ambiguous, partial), on matchConfidence = medium, or on leiLapsed / renewalOverdue.

Locally: put the input in <storage>/key_value_stores/default/INPUT.json, then CRAWLEE_STORAGE_DIR=$PWD/<storage> node src/main.js. Tests (no network): npm test.