Greenhouse · what the documentation says
What a Greenhouse application status actually tells you
Greenhouse documentation last checked on . Every factual claim below is sourced from Greenhouse’s own published documentation, linked at the end.
The short answer
Greenhouse documents that its AI Talent Matching is assistive and, in its own words, "does not automatically advance or reject candidates" — recruiters remain responsible for hiring decisions. It is also an add-on that an employer has to buy and switch on, with criteria the employer defines and weights themselves. Which statuses you see, and what they are called, is configured by the employer. So the widely-published tables of Greenhouse status meanings are not describing anything fixed.
What Greenhouse documents
Paraphrased from Greenhouse's own public documentation, not from third-party guides.
- AI matching does not auto-reject
- Greenhouse states that Talent Matching is assistive AI rather than automated decision-making, that it does not automatically advance or reject candidates, and that recruiters and hiring managers remain responsible for all hiring decisions. Its support site carries a question titled "Does Talent Matching auto-reject candidates with a low match score?" — the folklore is common enough that the vendor answers it directly.
- Matching is employer-calibrated, and optional
- Where it is in use, the hiring team first sets a calibration: criteria such as required skills, relevant experience and job titles, plus the importance assigned to each. Candidates are then sorted into categories such as Strong, Good, Partial, Limited or Needs manual review. It requires a paid add-on, and recruiters can override a candidate's match score. No outside tool can know an employer's calibration, which is precisely why an external match percentage is meaningless.
- There is a hard file-size limit
- Greenhouse Recruiting states it cannot parse resumes larger than 2.5 MB, and gives high-quality images as a common cause of exceeding it. This is a specific, checkable number, and it is smaller than most people assume.
- Placeholder-looking data gets skipped
- Greenhouse documents that resumes containing fake or poorly disguised data may parse only partially — a name like "First Last" or an obviously invented company name can be skipped by the parser while other fields populate. Relevant if you anonymise your resume for any reason.
- A failed parse is visible to the recruiter
- When parsing fails, Greenhouse shows that the resume was attached rather than parsed, and the recruiter has to enter the details by hand. A failed parse means friction for whoever is reading you, not a silent automatic rejection.
What the employer configures, not the software
This is the reason no page can hand you a table of status meanings, including this one.
- Whether Talent Matching is purchased and enabled at all, and the calibration criteria and weights if it is.
- Which status updates are shared with candidates, and how much of the internal pipeline is ever exposed.
- Whether a candidate-facing AI disclaimer appears on the job post, and whether candidates can request non-AI review.
- The stages, the wording, and the timing of any notification you receive.
What you can actually observe
- The updates the employer has chosen to share.
- Any AI disclaimer the employer has configured on the job post.
- Your own application and the files you attached.
What nobody outside the company can know
Including us. We would rather be useful about the limits than confident about things we cannot see.
- Whether this employer uses Talent Matching at all, and if so what criteria and weights they set.
- What match category you were placed in, or whether a recruiter overrode it.
- What a particular status wording means inside that company's process.
- Whether your resume parsed successfully on their side.
- We cannot see any of it either. That is why this page cites documentation instead of publishing a table of status meanings.
What you can actually test
Whether your resume file is machine-readable, and whether it trips the specific problems Greenhouse names in its own documentation. The 2.5 MB limit you can check yourself in seconds. Whether your text extracts cleanly, whether your dates and contact fields survive, and whether your layout leaves the fingerprints of a structure that flattens badly — that is what our extraction pass shows you.
Free, no account, and your resume is not saved. It is our own diagnostic, not Greenhouse’s parser — which is exactly why we only claim what it can observe.
Our view
Greenhouse publishes something most of this category will not admit: the AI is assistive, it does not auto-reject, and a human remains responsible. Which makes the mystery-percentage industry harder to justify, not easier. Get your resume readable, keep it under the documented size limit, and then spend your attention on the decision that actually moves your search: whether this role is worth applying to at all.
RoleSprint is a job-search product, so read that as a commercial opinion. It is also the argument we would make if we sold nothing: a readable resume is table stakes, and the decision that changes your search is which roles you spend your hours on.
And if several applications have gone quiet rather than just this one, the question is no longer about any single status page — read the roles you are chasing as one target market instead.
Sources
Greenhouse's own documentation. Not nofollowed — these are genuine citations supporting the claims above.