Data model

Every dataset's fields with their types and meanings, confidence and basis.

In short

An organization has a map of parts (components), each on a primitive of its industry's schema and in one of five layers (people, process, software, data, decisions). A part can carry figures (values). Every part and figure names its source, how sure we are (confidence) and what it stands on (basis). Organizations are linked to each other, identified in outside registers, and placed on the industry ontology.

The fields below are generated from the exporter's own column list, so they always match the files; enum values come from the database schema.

Confidence

How much a part or a figure can be trusted, on every row:

Value	Meaning
`verified`	Confirmed by the client, or read directly from inspected source material.
`self_reported`	Provided by the client, not independently checked.
`inferred_high`	Documented in the company's own public material.
`inferred_low`	Partial or indirect public evidence.

Basis: quoted or linked

What a row stands on, worked out from its source, its quote and where it came from. A quoted row has the sentence in quote and the page in source_url; a linked row has the page only.

Value	Meaning
`quoted`	A web page and the words copied from it. Research stores a finding only after checking the words are on the page.
`linked`	A web page without a quote: a structured record from a register, a filing or an import.
`self_reported`	The organization said it about itself (an owner's upload or publication). Labelled as theirs, never as verified.
`inferred`	Older research that kept a page but no quote. Read as inferred, never as fact.
`unsourced`	Nothing to open. Never on a public page, so rare in the files.

Organizations organizations

One public organization: name, website, kind, industry, region, claimed, counts. Open to everyone.

iduuid
The organization's id.
reftext
Its public reference: the name as a slug and the first 8 characters of the id (stripe-021b990f). API, MCP and CLI take it.
nametext
Its name.
websitetext
Its website.
domaintext
The registrable domain of the website (stripe.com).
kindenum
What kind of organization it is.
companygovernmentnonprofiteducationhealthcarecooperativeassociationotherinternationalreligiouspoliticalhousehold
industry_keytext
Where it sits in the industry ontology (a key in industries).
industry_nametext
That node's name.
regiontext
Where it is based, as its source states it.
size_bandtext
Its size band, when known.
business_modeltext
Its business model archetype (project, service, product, subscription, marketplace …).
claimedboolean
Whether the organization verified ownership of its profile.
wikidata_idtext
The Wikidata item it was imported from or matched to.
public_partsinteger
How many public parts its map has.
public_flowsinteger
How many public flows its map has.
updated_attimestamp
When a public part last changed (UTC).
profile_urltext
Its profile on systematlas.org.

Identifiers identifiers

One identifier of an organization in an outside register (Wikidata, LEI, CIK, EIN, ROR …). Open to everyone.

organization_iduuid
The organization.
schemeenum
The register (see Identifiers).
wikidataleicikeinrorueidotgoveutrswatchopenalex
valuetext
The identifier in its stored form.
source_urltext
Where the identifier was read.
upstreamtext
The upstream source the row comes from (a key in the sources catalog, web for the open web).
attributiontext
The attribution the upstream's license requires, when it requires one: keep it with the row.

Organization links organization_links

One stated relation between two organizations: direct or ultimate parent, funded by. Open to everyone.

from_organization_iduuid
The organization the relation is stated about.
to_organization_iduuid
The other organization.
relationenum
direct_parent and ultimate_parent (to is the parent of from, as GLEIF states it), funded_by (to funds from).
direct_parentultimate_parentfunded_by
source_urltext
Where the relation is stated.
quotetext
The words that state it.
as_ofdate
When the source stated it.
upstreamtext
The upstream source the row comes from (a key in the sources catalog, web for the open web).
attributiontext
The attribution the upstream's license requires, when it requires one: keep it with the row.

Industries industries

One node of the industry ontology, with its parent, standard and code. Open to everyone.

iduuid
The node's id.
keytext
Its stable key (naics_621210). Organizations point to it.
nametext
Its name.
parent_keytext
The node above it; empty for Universal, the root.
levelenum
Its depth: base (Universal), type (an organization type), sector, industry, segment (ours, refining a node).
basetypesectorindustrysegment
systemtext
The standard it comes from: naics, cofog, icnpo, sha, isced, … segment for our own.
codetext
Its code in that standard (621210, 07.3).
kindenum
The kind of every organization placed here or below, when the node sets one.
companygovernmentnonprofiteducationhealthcarecooperativeassociationotherinternationalreligiouspoliticalhousehold
descriptiontext
What the node covers.
aliasestext[]
What people call organizations here (; between them in CSV).
crosswalkjson
Its codes in other standards.
organizationsinteger
Public organizations placed directly on the node.

Parts components

One public part of an organization's map (a component), with its primitive, layer, confidence and source. Whole for Members; free accounts download slices.

iduuid
The part's id.
organization_iduuid
The organization (an id in organizations).
parent_iduuid
The part it sits under, when that part is public too.
primitive_keytext
The primitive of the industry schema it maps to; empty for a custom part.
labeltext
Its name on the map.
layerenum
The layer it belongs to.
peopleprocesssoftwaredatadecisions
kindenum
Internal to the organization, or an outside party it depends on.
internalexternal
external_kindenum
For an outside party: what kind.
customersupplierpartnerregulatoreconomic_factor
descriptiontext
What the part is, in a line.
confidenceenum
How sure we are: verified, self_reported, inferred_high (the organization's own public material), inferred_low (partial or indirect evidence).
verifiedself_reportedinferred_highinferred_low
basisenum
What the row stands on: quoted (a page and words copied from it), linked (a page, no quote: a structured record), self_reported (the organization said it), inferred (older research with a page and no quote), unsourced.
quotedlinkedself_reportedinferredunsourced
quotetext
The words on the source page it rests on (one sentence at most).
source_iduuid
The source (an id in sources).
source_urltext
The source page.
upstreamtext
The upstream source the row comes from (a key in the sources catalog, web for the open web).
attributiontext
The attribution the upstream's license requires, when it requires one: keep it with the row.
visibilityenum
public (research from public sources) or published (the organization published it).
publicpublished
provenanceenum
Where it came from: research, an import (registry), the organization itself, a third party.
company_uploadowner_publishedresearchthird_partyregistryoptimal_engine
review_statusenum
confirmed (reviewed and kept) or draft (found by research, not reviewed yet).
draftconfirmed
created_attimestamp
When it was recorded (UTC).
updated_attimestamp
When it last changed (UTC).

Figures values

One public figure on a part (revenue, employees, a count), with its period, confidence and source. Whole for Members; free accounts download slices.

iduuid
The figure's id.
organization_iduuid
The organization.
component_iduuid
The part it is stated on (an id in components).
keytext
What is measured: the parameter's key (revenue_usd, employees) or the figure's own key.
labeltext
The parameter's name, when the figure is on a schema parameter.
valuejson
The figure: a number, text or true/false (JSON in JSON lines, plain in CSV).
unittext
The parameter's unit, when it has one.
as_ofdate
The period the figure is stated for.
published_ondate
When its source published it.
confidenceenum
As on parts.
verifiedself_reportedinferred_highinferred_low
basisenum
What the row stands on: quoted (a page and words copied from it), linked (a page, no quote: a structured record), self_reported (the organization said it), inferred (older research with a page and no quote), unsourced.
quotedlinkedself_reportedinferredunsourced
quotetext
The words it rests on.
source_iduuid
The source.
source_urltext
The source page.
upstreamtext
The upstream source the row comes from (a key in the sources catalog, web for the open web).
attributiontext
The attribution the upstream's license requires, when it requires one: keep it with the row.
provenanceenum
As on parts.
company_uploadowner_publishedresearchthird_partyregistryoptimal_engine
statusenum
confirmed or draft, as on parts.
draftconfirmed
created_attimestamp
When it was recorded (UTC).

Sources sources

One source a public part or figure rests on: the page, its kind and when it was read. Whole for Members; free accounts download slices.

iduuid
The source's id.
kindenum
What kind of source.
websitecareersjob_postpressreviewfilingregistrymanualself_reported
titletext
Its title.
urltext
Its address.
retrieved_attimestamp
When it was read (UTC).
upstreamtext
The upstream source the row comes from (a key in the sources catalog, web for the open web).
attributiontext
The attribution the upstream's license requires, when it requires one: keep it with the row.

PreviousData overview Next Identifiers