How Openbound reaches a conclusion

Where the evidence comes from

Openbound reads two kinds of public record. The first is U.S. Department of Labor disclosure data: certified H-1B Labor Condition Applications, which show that an employer took the formal step of filing for a role, and PERM applications, which relate to permanent residence sponsorship. The second is the job posting itself, taken directly from the employer's own careers site.

These are separate signals and we keep them separate. A filing history describes what an employer has done before. A posting describes what this specific opening says today.

Matching a company to its filings

Employers rarely file under the name you see on a careers page. Filings arrive under legal entities — a company you know as Twitch appears as Twitch Interactive, and GoDaddy as GoDaddy.com, LLC. Openbound resolves a company to its filing entities and records how confident that match is.

Where a company cannot be matched to a verified entity, we say so rather than guessing. A confident match to the wrong employer would produce specific, credible, wrong advice — the single worst outcome for someone planning around a visa.

Whether they sponsor this role

An employer's overall filing count says very little about a particular opening. A company that has sponsored five thousand software engineers may never have sponsored a marketing manager, and treating those as the same fact is how sponsorship advice becomes misleading.

So role evidence is reported at three strengths. An exact match means the employer has filed for this same role. A related match means they have filed for adjacent work in the same discipline — nearby, but not the same claim. Where the filings we hold cover none of this kind of work, that is reported as a finding in its own right, not as missing data.

What the posting says

A posting that states a sponsorship restriction outranks every historical signal. An employer with a long filing record that says it cannot sponsor this role cannot sponsor this role, and the conclusion reflects that regardless of how strong the history is.

The reverse also holds: a posting that explicitly offers sponsorship is treated as decisive. Openbound quotes the phrase it matched so you can check the reading against the posting yourself.

Confidence is a separate question

The conclusion and our confidence in it are reported independently, because they answer different questions. “Likely, low confidence” is a real and useful result: it says the balance of evidence points one way while telling you not to plan around it.

Confidence reflects corroboration rather than volume. Several independent signals agreeing — role, location, recent activity, posting language — supports a conclusion more than one very large number on its own.

What public data cannot tell you

Disclosure data is filed on a delay and covers completed fiscal years, so the most recent months are always thinner than the rest. It records applications and certifications, not offers, and it says nothing about a team's budget, a hiring manager's discretion, or a policy that changed last quarter.

Many employers have no U.S. filing history at all. That is frequently correct rather than a gap in our coverage — plenty of companies genuinely do not sponsor — but it means an honest answer is often “we checked, and there is not enough here to call it”.

Openbound reports evidence. It does not predict decisions, and nothing here is legal advice.

Corrections

If a company is matched to the wrong employer, or a conclusion looks wrong against evidence you have, tell us at hello@theopenbound.com. Employer identity corrections are the most useful thing you can send.