Find a company's LEI
Find the LEI of a US-listed company from its ticker or CIK, or a whole list, through the EDGAR to GLEIF crosswalk, and see why a match by name goes wrong.
You have a company's ticker or its SEC CIK, or a spreadsheet column of them, and you need each company's Legal Entity Identifier (LEI), the 20-character code GLEIF's register gives a legal entity. Searching GLEIF by the company's name looks like the obvious route, and it is the one that goes wrong. This guide takes the route through the SEC's identifiers instead, on three of Meridian's open datasets.
For one company, type its ticker, CIK or name here. The lookup runs in your browser, against the published EDGAR ↔ GLEIF crosswalk.
Try , or .
The rest of the guide does the same in DuckDB, for one company and then for a list.
Before you start
You need DuckDB 2.0 or later. The data is read over the network from Meridian's open lake, and DuckDB reads only the parts of each file a query needs.
The guide uses three datasets:
- SEC EDGAR company tickers: each listed security's ticker, exchange and its company's CIK, the SEC's identifier for a filer.
- GLEIF legal entities: every entity in GLEIF's register, with its LEI, legal name and jurisdiction.
- The EDGAR ↔ GLEIF crosswalk: which LEI each CIK belongs to, with a tier saying how the link was made.
Step 1: Attach the catalogue
One ATTACH makes every dataset a table, and USE lets you name them without the meridian. prefix. Use anywhere explains the line.
INSTALL ducklake;
ATTACH 'ducklake:https://openlake.meridian.online/catalog/open.ducklake' AS meridian (READ_ONLY);
USE meridian;Step 2: See why matching by name goes wrong
The SEC and GLEIF write the same company's name differently. Here are five companies, looked up in GLEIF by the exact name EDGAR gives them:
SELECT s.ticker, s.company_name, g.legal_name, g.lei
FROM sec.company_tickers_exchange AS s
LEFT JOIN gleif.lei AS g ON lower(g.legal_name) = lower(s.company_name)
WHERE s.ticker IN ('AAPL', 'NVDA', 'MSFT', 'BA', 'KO')
ORDER BY s.ticker;Only Apple is found. EDGAR writes NVIDIA CORP and BOEING CO; GLEIF writes NVIDIA CORPORATION and THE BOEING COMPANY. These outputs were taken on 17 November 2026.
┌────────┬────────────────┬────────────┬──────────────────────┐
│ ticker │ company_name │ legal_name │ lei │
├────────┼────────────────┼────────────┼──────────────────────┤
│ AAPL │ Apple Inc. │ Apple Inc. │ HWUPKR0MPOU8FGXBT394 │
│ BA │ BOEING CO │ NULL │ NULL │
│ KO │ COCA COLA CO │ NULL │ NULL │
│ MSFT │ MICROSOFT CORP │ NULL │ NULL │
│ NVDA │ NVIDIA CORP │ NULL │ NULL │
└────────┴────────────────┴────────────┴──────────────────────┘Loosening the match to any name that contains the word finds plenty, and the company is lost among them. These are the first five, alphabetically, of the GLEIF entities whose names contain "nvidia":
SELECT legal_name, country, count(*) OVER () AS matches
FROM gleif.lei
WHERE legal_name ILIKE '%nvidia%'
ORDER BY legal_name
LIMIT 5;┌───────────────────────────────────────────────────────────────────────────┬─────────┬─────────┐
│ legal_name │ country │ matches │
├───────────────────────────────────────────────────────────────────────────┼─────────┼─────────┤
│ ASA STRATEGY AÇÕES NVIDIA CLASSE DE INVESTIMENTO EM AÇÕES - RESP LIMITADA │ BR │ 28 │
│ BMO Autocallable Income NVIDIA (NVDA) ETF – Series 1 │ CA │ 28 │
│ CHINAAMC NVIDIA DAILY (2X) LEVERAGED PRODUCT │ HK │ 28 │
│ CSOP NVIDIA DAILY MAX (-2X) INVERSE PRODUCT │ HK │ 28 │
│ CSOP NVIDIA DAILY MAX (2X) LEVERAGED PRODUCT │ HK │ 28 │
└───────────────────────────────────────────────────────────────────────────┴─────────┴─────────┘They are funds and leveraged products named after the company. Which row, if any, is the company has to be checked by hand, every time.
Step 3: Go through the CIK
The crosswalk links a CIK to an LEI. Each row carries a tier: authoritative where an official registration record links them, confirmed where the names match exactly once normalised (right about 95% of the time, the crosswalk's own measure), candidate where they only look alike, and ambiguous where one exact name matched more than one entity. Join EDGAR's tickers to it on the CIK:
SELECT s.ticker, x.lei, x.legal_name, x.tier, x.as_of
FROM sec.company_tickers_exchange AS s
JOIN library.edgar_lei AS x ON x.key_type = 'cik' AND x.key = CAST(s.cik AS VARCHAR)
WHERE s.ticker IN ('NVDA', 'AAPL')
ORDER BY s.ticker, x.tier;NVIDIA's row is confirmed. Apple has two: the confirmed Apple Inc. and a candidate, Apple-Metro, Inc., which is a different company. The crosswalk's as_of column dates each row.
┌────────┬──────────────────────┬────────────────────┬───────────┬────────────┐
│ ticker │ lei │ legal_name │ tier │ as_of │
├────────┼──────────────────────┼────────────────────┼───────────┼────────────┤
│ AAPL │ 5493005QKQO68Q3JGU22 │ Apple-Metro, Inc. │ candidate │ 2026-11-17 │
│ AAPL │ HWUPKR0MPOU8FGXBT394 │ Apple Inc. │ confirmed │ 2026-11-17 │
│ NVDA │ 549300S4KLFTLO7GSQ80 │ NVIDIA CORPORATION │ confirmed │ 2026-11-17 │
└────────┴──────────────────────┴────────────────────┴───────────┴────────────┘A candidate is a name guess, offered so that you can check it, and the crosswalk does not assert it, nor an ambiguous row. Keep only the two tiers it does assert.
Step 4: Look up a list
Put your tickers in a table and join them through the same two steps, keeping the asserted tiers. Each row of the result is a ticker and the LEI it reaches, or empty where the crosswalk has none:
WITH my_list(ticker) AS (
VALUES ('NVDA'), ('AAPL'), ('MSFT'), ('TSLA'), ('COIN'), ('BA'), ('KO'), ('DIS')
)
SELECT l.ticker, s.cik, x.lei, x.legal_name, x.tier
FROM my_list AS l
LEFT JOIN sec.company_tickers_exchange AS s ON s.ticker = l.ticker
LEFT JOIN library.edgar_lei AS x
ON x.key_type = 'cik'
AND x.key = CAST(s.cik AS VARCHAR)
AND x.tier IN ('authoritative', 'confirmed')
ORDER BY l.ticker;┌────────┬─────────┬──────────────────────┬───────────────────────┬───────────┐
│ ticker │ cik │ lei │ legal_name │ tier │
├────────┼─────────┼──────────────────────┼───────────────────────┼───────────┤
│ AAPL │ 320193 │ HWUPKR0MPOU8FGXBT394 │ Apple Inc. │ confirmed │
│ BA │ 12927 │ NULL │ NULL │ NULL │
│ COIN │ 1679788 │ 5493004G3J2SC154DU06 │ Coinbase Global, Inc. │ confirmed │
│ DIS │ 1744489 │ NULL │ NULL │ NULL │
│ KO │ 21344 │ NULL │ NULL │ NULL │
│ MSFT │ 789019 │ INR2EJN1ERAN0W5ZP974 │ MICROSOFT CORPORATION │ confirmed │
│ NVDA │ 1045810 │ 549300S4KLFTLO7GSQ80 │ NVIDIA CORPORATION │ confirmed │
│ TSLA │ 1318605 │ 54930043XZGB27CTOV49 │ TESLA, INC. │ confirmed │
└────────┴─────────┴──────────────────────┴───────────────────────┴───────────┘To run it on your own list, replace the VALUES with your file: SELECT upper(trim(ticker)) AS ticker FROM read_csv('your_tickers.csv'), since EDGAR writes tickers in capitals and a stray space would match nothing. With CIKs instead of tickers, join my_list straight to the crosswalk on x.key = CAST(cik AS VARCHAR) and leave EDGAR out. A CIK can reach more than one LEI, so a list can come back longer than it went in; check those rows by hand. A confirmed row is a name match, so for a list that matters, spot-check some of them against the jurisdiction in GLEIF, as Step 5 does.
Step 5: When the crosswalk has no answer
Boeing, Coca-Cola and Disney came back empty. The crosswalk's only rows for Boeing and Coca-Cola are candidates, and both are wrong (Boeing Way, LLC and a Greek Coca-Cola bottler), and it has no row for Disney. GLEIF writes all three with a leading "The" (THE BOEING COMPANY), which EDGAR leaves out.
For a company like these, look it up in GLEIF by its full legal name, as the company itself writes it:
SELECT legal_name, lei, jurisdiction, city
FROM gleif.lei
WHERE lower(legal_name) IN ('the boeing company', 'the coca-cola company', 'the walt disney company')
ORDER BY legal_name, jurisdiction;┌─────────────────────────┬──────────────────────┬──────────────┬────────────┐
│ legal_name │ lei │ jurisdiction │ city │
├─────────────────────────┼──────────────────────┼──────────────┼────────────┤
│ THE BOEING COMPANY │ RVHJWBXLJ1RFUBSY1F30 │ US-DE │ WILMINGTON │
│ THE COCA-COLA COMPANY │ UWJKFUJFZ02DKWI3RY53 │ US-DE │ WILMINGTON │
│ THE WALT DISNEY COMPANY │ 549300GZKULIZ0WOW665 │ US-DE │ WILMINGTON │
│ The Coca-Cola Company │ 8945007BLGJ517U74C54 │ AU │ NSW │
└─────────────────────────┴──────────────────────┴──────────────┴────────────┘Two entities are called The Coca-Cola Company, so the name alone does not settle it. The jurisdiction does: the company the SEC lists is incorporated in Delaware (US-DE), the state its annual report gives on its cover page. Check the jurisdiction every time you take an LEI from a name.
For one company, the GLEIF dataset page has a search by legal name or LEI that does the same in your browser.
How much the crosswalk covers
Across every company in EDGAR's ticker file, this counts how many reach an LEI at the two asserted tiers:
SELECT
count(DISTINCT s.cik) AS companies,
count(DISTINCT s.cik) FILTER (WHERE x.lei IS NOT NULL) AS with_an_lei
FROM sec.company_tickers_exchange AS s
LEFT JOIN library.edgar_lei AS x
ON x.key_type = 'cik'
AND x.key = CAST(s.cik AS VARCHAR)
AND x.tier IN ('authoritative', 'confirmed');┌───────────┬─────────────┐
│ companies │ with_an_lei │
├───────────┼─────────────┤
│ 7997 │ 4439 │
└───────────┴─────────────┘So a little over half reach one. The rest are companies with no LEI in GLEIF's register and companies, like Boeing, that the crosswalk does not link to theirs; this count does not tell the two apart. The data is republished as filings and registrations change, so your counts may differ from these.
Where to go next
- The crosswalk's page explains its tiers and methods, and lets you filter it in the browser.
- Use anywhere runs the same queries from Python, R or a script.
- The quick start is a tour of the datasets in your browser, from search to a downloaded CSV.
Use Anywhere
One ATTACH line exposes the whole catalogue to DuckDB 2.0 or later, from the command line, Python or R. Or read one Parquet file directly, or download it.
Data & Provenance
Where each Meridian dataset comes from, its open licence, how releases are published, its typed columns, what we promise about URLs, and the known limits.