What a debarred company's name starts with
Measured 2026-09-19 on 18787 company records from the U.S. federal exclusions list. There are 7198 different words those names begin with — and the ten most common cover 18.3% of the list. The most common opening is not a name at all: “limited liability”, on 1133 records.
Every guide to entity matching tells you to canonicalise legal-entity forms — “Ltd”, “GmbH”, “LLC”. None of them we could find says how much of a real list those forms account for, which is the number that decides whether canonicalising is tidy-up or the main event. On this list it is the main event.
The first word
| first word | records | of the list |
|---|---|---|
limited | 1142 | 6.1% |
joint | 730 | 3.9% |
aktsionernoe | 366 | 1.9% |
ao | 345 | 1.8% |
obshchestvo | 300 | 1.6% |
al | 168 | 0.9% |
fu | 123 | 0.7% |
public | 98 | 0.5% |
federal | 90 | 0.5% |
china | 81 | 0.4% |
open | 75 | 0.4% |
the | 66 | 0.4% |
The first two words, which is where it becomes obvious
| first two words | records | of the list |
|---|---|---|
limited liability | 1133 | 6% |
joint stock | 724 | 3.9% |
aktsionernoe obshchestvo | 349 | 1.9% |
obshchestvo s | 300 | 1.6% |
fu yuan | 123 | 0.7% |
public joint | 89 | 0.5% |
open joint | 74 | 0.4% |
federal state | 71 | 0.4% |
Roughly one company in eight on this list opens with a legal form rather than anything that
identifies it. limited liability, joint stock,
aktsionernoe obshchestvo and obshchestvo s — the last two being the Russian for
“joint-stock company” and the opening of “limited liability company” — are the four commonest openings on
the list, and none of them tells you which company you are looking at.
The tail is different in kind. fu yuan is a fishing-fleet naming convention;
federal state and public joint are more state-enterprise forms. The
concentration is not an artifact of one sanctions programme — it is what happens when a list
records companies under their registered legal names.
Why it matters if you are building matching
- Anything weighted toward the front of a name concentrates badly here. Prefix indexes, first-token blocking keys, “starts with” lookups — on this list, more than a thousand records share the same first two words, so a key built from them is not a key.
- Strip the form before you index, not after you match. Canonicalising at match time still leaves the blocking key useless. The counts above are exactly the list of openings worth skipping.
- Derive that list from your own corpus rather than copying ours. These counts are a property of this list on this date. A hand-written stoplist is only as current as the day it was typed; the two CSVs below are so you can compute your own.
- Non-English forms are most of the effect. Three of the top four openings are transliterated Russian. A stoplist of English legal forms would miss them entirely.
Method, scope and what is deliberately missing
-
18787 records, being those the source classifies
businessand dates on or after 2022-04-04. That cutoff is not arbitrary and not a sampling choice: SAM.gov's terms place company names on older exclusion records under a third-party data licence that forbids bulk use, so those records are not read, not counted, and not used to normalise anything. The finding is rebuilt from what is unambiguously publishable rather than reasoned around the restriction. - No company names are published, here or in the downloads. The files are token counts.
- Records the source classifies as individuals are not included — this service and this study handle organisation records only.
- Names are normalised before the first word is taken, with the same production matcher our screening APIs use, so “Limited Liability Company” and “LIMITED-LIABILITY COMPANY” count as one opening.
- One list, one date. A different list has different forms; a different date has different records. The shape is the finding, the exact percentages are of this snapshot.
Data
Source: SAM.gov Entity Exclusions, U.S. General Services Administration — a work of the U.S. Government. GSA is cited as the source and does not endorse this analysis.
Related: a company name without spaces misses the screening list, which is the failure this concentration makes worse, and the free screening name checker.
Corrections. If a figure here is wrong we correct it on this page with the date and what changed, and say so in the changelog.