SEC Subsidiaries & Major Customers Scraper (EDGAR) avatar

SEC Subsidiaries & Major Customers Scraper (EDGAR)

Pricing

Pay per event

Go to Apify Store
SEC Subsidiaries & Major Customers Scraper (EDGAR)

SEC Subsidiaries & Major Customers Scraper (EDGAR)

Subsidiary lists (Exhibit 21 / 20-F Exhibit 8) and 10%-customer/supplier concentration disclosures from SEC 10-K and 20-F filings, every row with a verbatim quote, document URL and accession number. Snapshot, year-over-year diff, and monitor modes. Deterministic parsing, no AI.

Pricing

Pay per event

Rating

0.0

(0)

Developer

Jonathan D Battles

Jonathan D Battles

Maintained by Community

Actor stats

0

Bookmarked

2

Total users

1

Monthly active users

a day ago

Last modified

Share

Turn U.S. public-company annual filings into a clean relationship graph: every subsidiary listed in Exhibit 21 (10-K) or Exhibit 8 (20-F), plus every major customer and major supplier disclosure (for example "Customer A accounted for 14% of revenue") with the exact quote, its position in the filing and a link to the source document.

By default it returns subsidiaries only. Major customer and supplier extraction is experimental and opt-in: add MAJOR_CUSTOMER and/or MAJOR_SUPPLIER to relationshipTypes to turn it on, and these links are incomplete: expect it to miss some customer and supplier disclosures (see Limitations).

Use it for supplier-risk screening, KYC/KYB ownership maps, credit and investment research, or to get an alert when a company adds or drops a subsidiary or a key customer.

What data does it extract?

One dataset row per relationship:

FieldExample
relationshipTypeSUBSIDIARY, MAJOR_CUSTOMER, MAJOR_SUPPLIER, NO_MAJOR_CUSTOMER, NO_MAJOR_SUPPLIER
sourceName / sourceTicker / sourceCikCummins Inc. / CMI / 0000026172
counterpartyName / counterpartyLabelsubsidiary or customer name as filed / Customer A, two customers
jurisdiction, hierarchyParentNameUnited Kingdom, parent entity when the exhibit is indented
metric, value, fiscalYear, thresholdPctPCT_REVENUE, 14, 2025 (also PCT_RECEIVABLES, PCT_PURCHASES, PCT_SEGMENT_REVENUE)
quote, quoteCharOffset, sectionthe sentence from the filing that supports the row
filingForm, accessionNumber, filingDate, documentUrl, filingIndexUrllinks straight to SEC EDGAR
extractionMethod, verificationhow the row was found and that its quote was checked against the filing text
edgeIdstable ID, so you can join runs

Every concentration row carries a verbatim quote that is re-checked against the filing text before it is saved. Rows that fail the check are dropped, never guessed.

How to use it

  1. Click Try for free and enter one or more tickers or CIK numbers (the prefill runs CMI).
  2. Pick a mode: snapshot, diff or monitor.
  3. Click Start. A single company usually finishes in under a minute.
  4. Download the results as JSON, CSV or Excel, or use the views Overview, Subsidiaries, Concentration and Changes.
  5. For alerts, schedule a monitor run and connect a webhook, Slack or email integration.

Modes

  • snapshot (default): the latest annual filing (or the fiscalYears you choose).
  • diff: latest vs. prior annual filing. Only change rows (SUBSIDIARY_ADDED, SUBSIDIARY_REMOVED, JURISDICTION_CHANGED, CUSTOMER_ADDED, CUSTOMER_REMOVED, SUPPLIER_ADDED, SUPPLIER_REMOVED, CONCENTRATION_CHANGED).
  • monitor: changes since your previous run of the same watchlistName. The first run stores a free baseline. Schedule it to get alerts.

Input

You only need companies. Leave secUserAgent empty to use the publisher's declared SEC contact, or enter your own.

{
"companies": ["CMI", "FFIV", "0001048695"],
"secUserAgent": "Your Name your.email@example.com",
"mode": "snapshot",
"relationshipTypes": ["SUBSIDIARY"],
"maxResults": 1000
}
  • secUserAgent is required. SEC fair-access rules ask every client to identify itself with a name and contact email. Enter your own. The actor sends it to sec.gov only.
  • companies: tickers or CIK numbers, up to 500 per run.
  • Optional settings: includeAnonymisedCounterparties, includeNegativeStatements, materialPctPoints, watchlistName, maxResults and maxTotalChargeUsd.

The run's OUTPUT record in the key-value store gives a status for each company (OK, NO_EX21_FOUND, BY_REFERENCE, NO_DISCLOSURE_FOUND, FETCH_FAILED, PARSE_FAILED, NO_ANNUAL_FILING, UNKNOWN_COMPANY, SKIPPED_BUDGET, ERROR, SEC_BLOCKED), so a missing exhibit is never silently confused with "no subsidiaries".

A problem with one company never ends the run: it is recorded in OUTPUT and the actor moves on. Rows are saved company by company, so if the run stops early (your spending cap, maxResults, the run's time limit, or SEC refusing requests) you keep everything delivered so far, and OUTPUT says why it stopped (stoppedByBudget, stoppedByMaxResults, stoppedByTimeLimit, stoppedBySecBlock).

How much does it cost?

Pay per event: you are charged only for results, never for compute.

  • company graph (one company processed, with at least one row): $0.02
  • customer or supplier relationship row that names the counterparty or gives an exact percentage: $0.003
  • change row (diff / monitor): $0.03
  • Free: subsidiary rows within a company graph, "no major customer/supplier" rows, rows that only say "N customers each exceeded 10%" (no name, no exact figure), and the monitor baseline.

Output example

{
"sourceCik": "0000026172",
"sourceName": "CUMMINS INC",
"sourceTicker": "CMI",
"filingForm": "10-K",
"accessionNumber": "0000026172-26-000009",
"filingDate": "2026-02-10",
"periodOfReport": "2025-12-31",
"filingIndexUrl": "https://www.sec.gov/Archives/edgar/data/26172/000002617226000009/0000026172-26-000009-index.htm",
"retrievedAt": "2026-10-12T03:52:06+00:00",
"attribution": "Source: U.S. SEC EDGAR filings (public information)",
"edgeId": "32f9415b9fc7f398b191",
"relationshipType": "MAJOR_CUSTOMER",
"counterpartyName": null,
"counterpartyLabel": "largest customer",
"counterpartyNamed": false,
"counterpartyNameNormalised": null,
"jurisdiction": null,
"doingBusinessAs": [],
"hierarchyParentName": null,
"notes": null,
"metric": "PCT_REVENUE",
"value": 13,
"fiscalYear": 2025,
"aggregateOnly": true,
"thresholdPct": null,
"noOtherAbove10Pct": false,
"documentUrl": "https://www.sec.gov/Archives/edgar/data/26172/000002617226000009/cmi-20251231.htm",
"section": "ITEM 1. Business",
"quote": "PACCAR is our largest customer, accounting for 13 percent of our consolidated net sales in 2025, 16 percent in 2024 and 16 percent in 2023.",
"quoteCharOffset": 41531,
"extractionMethod": "DETERMINISTIC_TEXT",
"confidence": 1,
"verification": "QUOTE_VERIFIED"
}

(Illustrative values; real rows always link the exact filing.)

Limitations

  • Data comes only from 10-K and 20-F filings and their Exhibit 21 / Exhibit 8. No 10-Q, 8-K or proxy data.
  • Exhibit 21 may legally omit subsidiaries that are not "significant", so the list can be incomplete.
  • Extraction is rule-based and deterministic (no AI guessing). Unusual layouts may yield a status such as NO_DISCLOSURE_FOUND instead of rows.
  • Companies often anonymise customers ("Customer A"). The actor keeps the label as filed and does not try to identify them.
  • A customer name is filled in only when the filing states it next to the figure ("PACCAR is our largest customer, accounting for 13 percent...", "sales to The Home Depot were 11%...", or a named-customer table). On a test set of 9 recent 10-Ks the actor never put a wrong name on a figure, and it found 29 of the 43 named figures. It still misses some phrasings, such as "39% of net sales were to Airbus" or names that read like a person's (for example "Lockheed Martin" in a bare table). Those figures are kept with an anonymous label instead.
  • A combined figure such as "sales to The Home Depot and Lowe's were 28% of the segment" is never split or attributed to one company.
  • No LEI, ownership percentages or corporate-registry enrichment.
  • Requests to sec.gov are kept under 10 per second, as SEC's fair-access policy requires.

Source: U.S. SEC EDGAR filings (public information). This actor is not affiliated with or endorsed by the SEC. It is not investment advice.

  • UK Late-Payer & AP-Prospect Signals (Battles_2026/uk-late-payer-signals): payment-practice and late-payment data for UK companies.

FAQ

Is it legal to use this data? Yes. SEC EDGAR filings are public records published by the U.S. government for free. The actor follows SEC's fair-access policy: it declares a User-Agent with a contact email on every request and stays under 10 requests per second. It reads only public filings and logs in nowhere. You are responsible for how you use the data; consult a lawyer if unsure.

Why are some subsidiaries missing? Exhibit 21 may legally omit subsidiaries that are not "significant". The per-company status in OUTPUT tells you what was found.

Do I pay for compute? No. You pay only the per-event prices above. Subsidiary rows and the monitor baseline are free. Set maxTotalChargeUsd to cap spend.

Can it name anonymised customers? No. It keeps "Customer A" exactly as filed and never guesses.

Can I get alerts? Yes: schedule monitor mode and connect an integration.