The Visa Sponsorship Signal Decoder, Posting by Posting

You need work-visa sponsorship, and you are looking at a single posting, deciding whether to spend an application on it. This is the reference for that decision: it maps every sponsorship signal - the posting phrase, the application-form question, the USCIS petition record, the DOL LCA filing, and E-Verify status - to what it proves and how it misleads, then tells you which source overrides which. Read it once to learn the routing, then jump to the row you need mid-search.

The signal that trips everyone is the heavy sponsor whose postings say "no sponsorship." I resolve that here by naming the precedence order plainly, so you route a posting to apply, ask, or skip in under a minute instead of applying blind.

Why the posting is the weakest signal you have

The posting is the signal you weight most and the one that lies most. A posting is a recruiter-owned template; a filing is a legally attested act. When they disagree, the filing wins.

The documented case makes this concrete. Department of Labor data shows one employer sponsoring roughly 500 H-1Bs a year while nearly all of that employer's job postings say it does not sponsor. The cheap, un-vetted artifact contradicts the expensive, audited one. If you had read only the posting, you would have skipped a company filing 500 petitions annually.

Silence compounds the problem, because silence is the normal state, not the exception. In one dated scan of a single job board, most posts said nothing about sponsorship at all. That is not 61.5% of employers refusing. It is 61.5% of postings where the topic was omitted, because sponsoring costs legal fees and lottery risk, so most employers commit to neither answer in the posting. A silent posting therefore carries almost no information. It has to be resolved by the filing record.

The cheap, un-vetted artifact contradicts the expensive, audited one, and the audited one wins.

The five signals and what each one proves

There are five signals in play for any posting. Each proves something narrow, and each misleads in a specific way. Learn the "proves" and the "lies like this" for every row and you will stop over-reading the posting.

Two rows deserve emphasis. The application-form question is not employer intent for the role. By law, the only two immigration questions askable of an applicant are "Are you legally authorized to work in the United States for any employer?" and "Will you now or will you in the future require employment visa sponsorship?" These are auto-populated by the ATS or the job board, standardized across postings, and written by no one at the company. The checkbox tells you nothing about whether this req is sponsorable.

The E-Verify row is the false friend. E-Verify is I-9 work-authorization checking, a different program from sponsorship. E-Verify employers are different from visa-sponsoring companies. The search shows account status and enrollment dates, nothing more. An enrolled twelve-person firm that never filed an H-1B will still appear as "enrolled."

Classify the posting language into four buckets

Every posting sorts into exactly one of four buckets: available, explicit exclusion, silent, or trap. Exclusion language is reliable. The other three are not what they look like.

• Available. Direct statements: "visa sponsorship is available," "we sponsor H-1B transfers," "open to all visa types," "H-1B transfer accepted," "EAD holders welcome." This is a real yes. Apply.
• Explicit exclusion. "Unable to provide visa sponsorship" is the most common phrasing, with several equivalents like "must have existing right to work" and "no sponsorship offered." This is a real no for the posting, though the filing record can still make it worth a recruiter question.
• Silent. No mention either way. This is the majority state and the one you must resolve against filings.
• Trap. "Must be authorized to work in the United States." This looks like exclusion and is not. It means legal permission on Day 1. OPT, an active H-1B, or an EAD already meets it, and many such companies still sponsor H-1B transfers and green cards.

The trap phrase is the second-most-common self-rejection after treating silence as "no." An OPT or EAD holder reads "must be authorized to work" and closes the tab. Do not. Route it as silent, then check the filings.

Here is the full stance breakdown from one dated board scan. Treat the exact percentages as a single dated data point from a single UK-tilted vendor, not a universal figure. The shape is what matters: available is tiny, silent is the majority.

The direction is corroborated elsewhere: the share of one large board's postings mentioning visa or green-card support rose by almost 285% by October 2024 versus the pre-pandemic baseline. Sponsorship visibility is rising, but it started from a low base, so most postings still say nothing.

The public sources of actual sponsorship history

Four public sources hold real sponsorship history, and they sit in a precedence order. USCIS and DOL data are legally attested filings and override the posting. E-Verify and the posting field are the weak signals.

USCIS H-1B Employer Data Hub. Covers FY2009 through FY2026 Q3 on employers who submitted H-1B petitions, queryable by fiscal year, employer name, city, state, zip, and NAICS code. It identifies employers by the last four digits of their tax identification. Its limits matter: it reflects first decisions only, excluding appeals, revocations, and pending cases; it refreshes on a lag; and it does not publish wage amounts, denial reasons, or a key tying a petition to a specific LCA. Use it for the yes/no question of whether an employer has sponsored, and the rough scale.

DOL OFLC disclosure data. Public disclosure through FY2026 Q3 for PERM, LCA (H-1B, H-1B1, E-3), H-2A, H-2B, CW-1, and Prevailing Wage, released quarterly. The LCA files are the richest source you have: each record carries 75+ columns, including job title, SOC code, worksite, and wage level. The LCA is Form ETA-9035, filed before the petition. This is where you confirm that an employer attested to roles like yours, not just that they filed something.

E-Verify Employer Search. Shows employer name, DBA, account status (enrolled or terminated), enrollment date, and termination date. The tool lists only employers self-reporting five or more employees. Enrollment is not sponsorship. Full stop.

PERM. Sits inside the same OFLC files and signals green-card intent. A recent PERM record for a role like yours is a strong sign the employer takes sponsored hires all the way to permanent residence.

Route a posting in under a minute: the procedure

This is the eight-step routing you run per posting. The first seven are the read; the eighth happens on the recruiter call. The whole read takes under two minutes once the sources are familiar, and often under one.

The routing table at the end of the read is short:

• Explicit exclusion and no filings for similar roles: skip.
• Filings present but posting silent or trap: apply, or ask the recruiter first if you want to conserve applications.
• Explicit availability: apply.
• Explicit exclusion but heavy recent filings for roles like yours: ask the recruiter, because the exclusion may be a template default.

If running the Data Hub and the LCA files for every posting sounds like a lot of manual lookup across a wide search, that is the friction Refolk removes. I can surface employers by their filing history and role, so you start from a filtered list of companies that have actually sponsored for work like yours rather than reading postings blind.

How this read goes wrong

The routing fails in predictable ways, and each failure has a specific check. This is the part of the reference to reread when a decision feels off.

Treating silence as "no." The most common and most expensive error. You skip a heavy filer. Silence is not a no; it often means the recruiter did not include it or the template lacked the section. Check: run the name through the Data Hub before skipping.

Reading "must be authorized to work" as exclusion. OPT, EAD, and current-H-1B holders self-reject on the trap phrase. Check: remember that OPT, an active H-1B, or an EAD already meets Day 1 authorization, and many such firms sponsor transfers. Route it as silent.

Trusting the ATS form field as employer intent. The Yes/No question is boilerplate, standardized by law and auto-populated by the board. Check: verify against filings, not the checkbox.

Assuming E-Verify equals sponsor. Enrolled small firms that never file H-1Bs will show as enrolled. Check: E-Verify employers are different from visa-sponsoring companies; confirm with the Data Hub or LCA files.

Assuming a heavy filer sponsors every role. Even the employer filing 500 a year can gate sponsorship internally. One report notes sponsorship there required SVP approval, so a "no" on a specific req can be genuine despite the filing volume. Check: ask the recruiter about this specific req.

Name-matching errors in the Data Hub. Employers appear under legal names and subsidiaries, and the Hub shows only the last four tax digits. Check: try parent, DBA, and subsidiary names before concluding there is no history.

Comparing an incomplete fiscal year. FY2026 counts are partial and were suppressed by a funding lapse. Check: do not read a low current-year count as declining sponsorship; weight complete prior years.

Answering the form inconsistently across boards. Select Yes in one ATS and No in another and you risk being flagged as inconsistent. Check: answer the same way everywhere. An H-1B transfer counts as needing sponsorship, so select Yes if the company must commence a sponsorship case now or in the future.

Your visa category rewrites the same posting

The same silent posting is a different read depending on your category, because category changes the employer's cost and uncertainty. Capless, treaty-based routes make silence a much softer "no" than it is for a cap-subject H-1B candidate.

For a cap-subject H-1B candidate, the employer faces legal fees plus the lottery. That uncertainty is exactly why so many postings stay silent, and it is why a silent posting is a genuine question mark. But the lottery odds are better than the raw applicant flood suggests. The selection rate rose from 24.8% in FY2024 to 35.3% in FY2026, while eligible registrations fell 26.9% between FY2025 and FY2026 as the beneficiary-centric rule cut duplicate registrations.

For a TN, E-3, or H-1B1 candidate, the employer's main objection largely evaporates. The E-3, for Australians, has no annual cap and no lottery, with two-year terms renewable indefinitely. TN covers listed treaty occupations - engineers, accountants, scientists, economists, analysts - for Canadians and Mexicans. H-1B1 serves Singaporeans and Chileans. These routes are cheaper and more predictable for the employer, so a silent posting is a weak "no" for you. The O-1, which requires meeting three of eight criteria and carries dual intent, and the L-1 intracompany transfer, tied to the sponsoring employer, sit outside the cap logic too.

There is a reason you must read the employer's filings rather than infer sponsorship from other candidates' profiles: candidates hide their status. In Refolk's index of professional profiles, entry-level "H1B" and entry-level "OPT/F1 STEM" headline cohorts in the US return effectively zero, meaning candidates rarely advertise visa status in a headline. The senior professionals who do surface "H1B visa sponsorship" in a headline are a very small, staffing-heavy cohort clustered at IT staffing and consulting firms. Applicants suppress status to avoid being screened out, which is precisely why the employer's filing record is your only reliable lens.

The done check and how to keep it current

Before you tag a posting, run this checklist. If every box holds, your tag is trustworthy.

To keep this reference current, re-check the mechanisms, not the numbers. The stance percentages come from a single dated board scan and will drift; treat them as shape, not fact. The Data Hub and OFLC files refresh quarterly and lag, so a low current-year count is not a decline. The cap selection rate moves each fiscal year with registration volume and the beneficiary-centric rule, so pull the latest published rate before you weight a cap-subject candidate's odds. And re-read the two application-form questions each cycle only to confirm they are still the standardized, legally bounded pair. The precedence order - filings over posting text, recruiter over filings for the specific role - is the durable part. Bookmark that, and the routing survives every refresh.

When you are running this across dozens of companies at once, the bottleneck is not the routing logic, it is the lookup volume. Starting from a list of proven sponsors for your exact role, rather than reading postings blind, is what turns a week of blind applications into a filtered batch. That is the friction I built Refolk to remove.