Skip to content

Research · Dated snapshot

The Sponsorship Signal Index: do job postings say whether they sponsor visas?

By · Updated · Postings read 29 September 2026, 10:20 UTC

The answer, September 2026

In a September 2026 location-screened sample of 11,304 US job postings from 2,000 employers, 94.0% said nothing about visa sponsorship, 5.9% said they would not sponsor or required US citizenship, and 0.1% said they would.

In the location-screened US internship subset of 984 postings, 17.3% refused sponsorship and 0.4% offered it.

Even at employers with an approved H-1B petition or a certified labor condition application on file, 15.7% of screened US internship postings still refused, so a company's filing history does not tell you whether a specific role sponsors.

US postings silent on sponsorship
94.0%
US internship postings that refuse
17.3%
Internships that refuse at H-1B or LCA filers
15.7%

Free to reuse under CC BY 4.0: download the September 2026 CSV (12,242 rows, one per posting, employer names and posting links included). Please link this page when you cite it.

What US job postings say about sponsorship.

Each posting falls in one of three groups. It refuses if it says sponsorship is not available or requires US citizenship or US-person status. It offers if it says sponsorship is available. Otherwise it is silent.

Share of postings that are silent, refuse or offer sponsorship, for the screened US posting sample and screened US internships. The same numbers are in the table below.

All US postings (sample) n=11,304

Refuses 5.9%Offers 0.1%Silent 94.0%

US internships (location screened) n=984

Refuses 17.3%Offers 0.4%Silent 82.3%

Share of postings by what they say about sponsorship, September 2026.
PopulationPostingsSilentRefusesOffers
All US postings (sample)11,30494.0%5.9%0.1%
US internships (location screened)98482.3%17.3%0.4%

Most sampled employers never say.

Posting shares can be moved by a few employers that post many roles, so the index also counts employers. An employer is silent here if none of its postings in the population said anything about sponsorship.

Share of employers by what their postings say. An employer with both a refusing and an offering posting counts in both columns.
PopulationEmployersSilent on every postingAt least one refusalAt least one offer
All US postings (sample)2,00089.0%10.7%0.3%
US internships (location screened)25084.8%14.8%0.4%

Refusals are concentrated. Two employers account for 51.8% of refusing US internship postings this month, and 9.6% of refusing postings in the all-postings sample. That is why the employer shares sit beside the posting shares every month.

A filing record is not a promise.

Many students check whether a company has filed for H-1B workers before applying. A past petition approval or certified labor condition application is historical evidence; it does not establish that the employer will sponsor this role. At employers with a filing record, some postings still refuse.

Share of US postings that refuse sponsorship, at employers with a filing record compared with all employers.
PopulationRefuse, at filing employersRefuse, all employersFiling employers with at least one refusal
US internships (location screened)15.7% (53 of 338)17.3%8 of 68 (11.8%)
All US postings (sample)6.3% (164 of 2,607)5.9%54 of 430 (12.6%)

A filing record here means an approved H-1B petition in the USCIS H-1B Employer Data Hub, a certified labor condition application (LCA) in the Department of Labor's disclosure data, or both. An LCA is a step before a petition, not an approval.

US postings at filing employers by type of record (all-postings sample, n=2,607).
Record on filePostingsShare
Approved petition and certified LCA1,77067.9%
Approved petition only41415.9%
Certified LCA only42316.2%

Pay listed on the posting.

A side measure from the same postings. It covers Greenhouse boards only, because pay from the other platforms on the feed is not captured, so no all-platform figure is claimed.

Share of US Greenhouse postings that list pay.
PopulationGreenhouse postingsList pay
All US postings (sample)4,68329.3% (1,373)
US internships (location screened)65017.7% (115)

How to read a posting for sponsorship.

Because most postings say nothing, the posting text is the first check, not the last. Career offices can link this list from their OPT and CPT pages.

  • Search the posting for "sponsor", "visa", "citizen", "authorized to work" and "US person".
  • "Must be authorized to work without sponsorship now or in the future" is a refusal, even when the word "visa" never appears.
  • A US citizenship or US-person requirement, often tied to export controls or a security clearance, rules out most visa holders.
  • Silence is not a no. Ask the recruiter, and answer the application's sponsorship question honestly. How to answer the sponsorship question.
  • A company's H-1B filings are a good sign, not a promise for this role. The table above shows refusals at filing employers.

To start from roles that are more likely to sponsor, the Litos board keeps postings that offer sponsorship or come from employers with H-1B filings on record: internships at employers that sponsor, or all jobs with the sponsorship filter.

Method and limits.

Litos reads postings from public employer boards and serves them through its jobs API. On 29 September 2026, the pull read the paginated internship feed and every sixth page of 100 from the API's ordered US list. The API labelled 1,000 internships and 68,467 postings as US at that time. We then required the visible posting location to identify a US worksite. That left 984 internship postings from 250 employers and 11,304 sampled postings from 2,000 employers. These are location-screened subsets of that day's feed, not a count of every US job market posting.

The location screen excluded 1 foreign and 15 ambiguous internship candidates, plus 35 foreign and 61 ambiguous sampled candidates. The CSV keeps their source rows with both original candidate flags and final cohort flags, so the exclusions and denominators can be checked. A posting counts once if it appears in both populations. The original candidate CSV is preserved in Git history with SHA-256 66f1eed211fbc17dc5e7f9852c5a7524ecd160f30c2be907e5974cbb76fed92f.

Visible locations use the frozen study location screen, version study-visible-location-v2-2026-09-29. Its location tables started from Litos's backend rule, but this study applies stricter checks to avoid counting a parent-office country or a foreign-only location as a US worksite. A mixed US and foreign location remains eligible only when it explicitly names a US worksite. Locations such as "Remote" without a US place are excluded as ambiguous.

The API assigned each posting's sponsorship status from its full text at collection time: it refuses if it says sponsorship is unavailable or requires US citizenship or US-person status, offers if it says sponsorship is available, and is silent otherwise. A refusal anywhere in the posting wins over an offer. The downloaded CSV reproduces the published cohort counts and rates, plus the visible-location exclusions. It does not contain the full posting text, so it cannot independently recreate the API's historical classification or the original API totals. Each row gives the source posting URL and read date; posting text may since have changed.

Employer filing records come from the USCIS H-1B Employer Data Hub (approved petitions, fiscal years 2021 to 2023, the years in Litos's employer-matching data) and the Department of Labor's fiscal year 2025 LCA disclosure files (certified applications), matched by legal name with hand-checked aliases.

Limits

  • The feed covers the boards Litos reads, such as Greenhouse, Workable, Breezy, Recruitee and Ashby. Workday and iCIMS postings are not on it, so large enterprise employers are under-represented.
  • The rule is conservative: "we sponsor, but not for every role" counts as a refusal.
  • Filing matches are strict, so the filing-employer counts are a lower bound.
  • This is a snapshot of one day. Postings change and close daily. The US sample was drawn from pages ordered by the API, so it should not be treated as a random sample of the whole US job market.
  • The September source pull did not archive all raw internship API rows, so its pagination and non-US count cannot be independently replayed from the CSV. The CSV does preserve all candidates included in the original published US cohorts and each visible-location exclusion.

Download the data.

September 2026 CSV, one row per posting, under CC BY 4.0. You may reuse it for any purpose, including commercially, with credit and a link to this page.

Suggested citation: Litos, "Sponsorship Signal Index, September 2026", trylitos.com/research/sponsorship-signal-index, data read 29 September 2026.

Columns in the CSV.
ColumnMeaning
read_onThe date the posting was read
employer, title, locationAs the employer's board shows them, with any em dash written as a hyphen
employment_typeInternship, Full-time and so on, when the board states it
atsThe applicant tracking system hosting the posting
sponsorship_statussilent, refuses or offers, by the rule above
employer_filing_recordboth, uscis_h1b (approved petition), dol_lca (certified LCA) or none
pay_listedyes if the posting states a salary figure
first_seen_onThe date Litos first saw the posting
posting_urlThe posting on the employer's board
candidate_us_postings_sample, candidate_us_internshipWhich original API-US draw contained the posting before checking its visible location
visible_location_countryUS, non-US or unknown from the frozen location rule
in_us_postings_sample, in_us_internship_subsetWhich location-screened population the row counts in; a posting can be in both

Wording on a posting can change after the read date. Check the posting itself before quoting any one employer.

Month by month.

Future snapshots will be added here with each month's file kept at its own address.

Share of postings that are silent, refuse and offer, by release.
MonthUS postings: silent, refuse, offerUS internships: silent, refuse, offerData
September 202694.0%, 5.9%, 0.1%82.3%, 17.3%, 0.4%2026-09.csv

Questions.

Do most job postings say whether they sponsor visas?

No. In the September 2026 location-screened index, 94.0% of sampled US job postings said nothing about visa sponsorship either way, and 89.0% of employers were silent on every sampled posting. Among screened US internships, 82.3% were still silent.

If a company files H-1B petitions, will it sponsor me?

Not necessarily. At employers with an approved H-1B petition or a certified labor condition application on record, 15.7% of screened US internship postings (53 of 338) still said the role would not sponsor or required US citizenship. A past petition approval or certified labor condition application is historical evidence; it does not establish that the employer will sponsor this role.

How do I find jobs that sponsor?

Read each posting for a sentence about sponsorship or citizenship before you apply, and ask the recruiter when it says nothing. The Litos jobs board has a filter, Only jobs where the company sponsors a work visa, that keeps postings which offer sponsorship or come from employers with H-1B filings on record. A filing record is historical evidence, not a promise for a particular role.

Sources.

Changelog.

  • 2026-09.csv. First release. A visible-location screen excludes foreign-only and ambiguous rows from the API-US draw. The CSV retains candidate flags and excluded rows for audit.

Keep going.

Written by , founder of Litos. Questions about the data: contact.