How Visa Grep works
Where our data comes from, how we match it to a company, and how the Sponsorship score is worked out.
Where the data comes from
- H-1B filings (LCAs) and green card filings (PERM): the U.S. Department of Labor's Office of Foreign Labor Certification disclosure data, FY2020 through June 30, 2026.
- H-1B approvals: the USCIS H-1B Employer Data Hub. FY2022 is the last complete year; FY2023 covers only part of the year, and USCIS hasn't published employer-level data since.
- E-Verify: the list of E-Verify employers USCIS published for December 31, 2018, the only one it has published.
Fiscal years run October to September. Every number on a company page is read from one row of our copy of these files, and the page lists those rows under Sources.
How filings are matched to a company
Filings carry the employer's name as typed on each form, so one company appears under many spellings. We group filings by a normalized form of the name and show the most common spelling as the company's name.
Employer IDs (FEINs) are shown but never used to merge companies: one ID can cover many differently named entities, and one company can use several IDs. Companies that share an Employer ID are listed as related, never combined.
USCIS approvals are linked to a company by name. Where USCIS publishes the last four digits of the Employer ID, we check that they agree.
How the Sponsorship score works
The score adds up four parts, each with a published rule, for up to 10 points. A part without enough data isn't rated, and the maximum shrinks rather than counting it as zero, so a page can show 8/10 or 5/8. The score describes the past record; it doesn't rate the company or predict any application.
| Part | Points | Rule |
|---|---|---|
| Recent approvals | 0–3 | initial approvals in FY2022: 0 -> 0, 1-4 -> 1, 5-24 -> 2, 25+ -> 3; not rated when the Hub FY2022 export does not list the employer under this name |
| Entry-level hiring | 0–3 | Level I+II share of certified H-1B cases with a level, FY2023-FY2025, at least 5 such cases: under 15% -> 0, 15-35% -> 1, 35-60% -> 2, 60%+ -> 3; fewer than 5 -> not rated |
| Files every year | 0–2 | fiscal years of FY2023-FY2025 with a certified H-1B case: 3 -> 2, 2 -> 1, 0-1 -> 0 |
| Approval rate | 0–2 | initial approval rate FY2020-FY2022 with at least 10 initial decisions: 95%+ -> 2, 85-95% -> 1, under 85% -> 0; fewer than 10 -> not rated |
Not enough data: a company with fewer than 5 certified H-1B filings in FY2023–FY2025 and fewer than 5 new petitions approved in FY2022 gets no score at all. That's its own state, not a low score: a company with four filings doesn't score 4/10.
How we work this out
scored | insufficient. `insufficient` means fewer than five certified H-1B LCA cases in FY2023-FY2025 AND fewer than five FY2022 initial approvals, so every component is NULL and the counts on this row are all there is to say. It does not mean the employer scored low, and it does not mean the employer does not sponsor.
What each flag means
These facts sit beside the score and are never part of it. Under each, “How we work this out” opens the rule as our database states it.
- H-1B dependent
Whether the company says, on its H-1B filings, that H-1B workers make up a large share of its staff. Dependent employers make extra promises about recruiting and not displacing U.S. workers.
How we work this out
The employer answered Yes to H1B_DEPENDENT on at least half of the certified LCAs where it answered the question, in FY2023-FY2025. false means it answered and Yes was under half; NULL means no certified case answered the question (E-3 and H-1B1 filings answer N/A). The floor is there because a single Yes among thousands of No answers is a tick in the wrong box: it would badge 979 employers that said Yes once, one of them on 1 filing of 2,760; the 473 that answered only once or twice still are. A dependent employer is not a worse employer: the answer is a statutory test on the share of H-1B workers on staff, not a finding.
- Willful violator
Whether the company says, on its filings, that DOL has found it to be a willful violator in the five years before. We report its own answers, which aren't checked against DOL's records.
How we work this out
The employer answered Yes to WILLFUL_VIOLATOR on at least two LCAs in FY2020-FY2025, certified or not, in at least two fiscal years, and in at least one of those years Yes was half or more of what it answered. The floor is there because a single Yes among thousands of No answers is a tick in the wrong box: 505 employers answer Yes exactly once and never twice, and the two employers with the most Yes answers say it on one filing in two thousand. false means the employer answered and did not meet that bar; NULL means it never answered. THIS IS THE FILER'S OWN ATTESTATION ON THE LCA, NOT A DOL FINDING. The authoritative source is the DOL published debarment list, which this project does not yet mirror - it is a follow-up catalog entry - so nothing here has been checked against it.
- Files for green cards
Whether the company filed any green card applications (PERM) recently. It says nothing about whether it would file one for any particular worker.
How we work this out
The employer has at least one PERM case decided in FY2023-FY2025, of any status. It says the employer files permanent-residence cases; it does not say any of them were certified, and it does not say the employer would file one for any particular worker. The per-status counts are in employer_years.
- Possibly cap-exempt
An indicator from the company's name or industry code that it may be a college or university, which can file H-1B petitions outside the annual cap. Only USCIS decides.
How we work this out
INFERRED, not a determination. true when an FY2023+ LCA carries a NAICS code beginning 6112 or 6113 (junior colleges, colleges and universities) or a raw spelling of the employer's name contains the word UNIVERSITY or COLLEGE. Never false: the column is NULL when neither signal is present, because the absence of a signal is not evidence that an employer is cap-subject. Non-profit research organisations and affiliated hospitals are cap-exempt and both tests miss them. Only USCIS decides cap exemption, petition by petition.
- E-Verify
Whether the company is on USCIS's list of E-Verify employers dated December 31, 2018. Not being on it tells you nothing.
How we work this out
The employer's name key appears in the USCIS E-Verify participating-employers list with a state the employer files in. That list is one snapshot, enrolled as of 2018-12-31, so false means only "not on the 2018 list" - not that the employer does not use E-Verify today, and not that it did not enrol the following week. The list carries no FEIN and no case number, so the match is by normalised name and state alone.
What we don't do
- We don't rank companies: a list sorted by filing volume would be a staffing-firm leaderboard, not an answer.
- We don't predict outcomes: no filing history can tell you whether your petition will be approved.
- We don't blend the agencies: DOL filings and USCIS approvals count different things and are never added together.
Which companies have a page
The 10,000 companies with the most Department of Labor filings (H-1B and green card filings together) have a page, covering 79.3% of all 4,674,037 filings on record. We don't know which companies people search for most, because we don't track searches, so filing volume decides which companies get a page. It never decides an order: the list is alphabetical.