Suspect an automated hiring screen? Check the application record for an observable issue before deciding what caused the rejection.
A standard rejection message does not tell you whether a person, an eligibility rule, a parser, an assessment, or another step determined the outcome. The U.S. EEOC says some employers use software, including AI, to score applications or tests and provides worker guidance on disability-related concerns. That does not let an applicant infer what happened in one case. The useful next step is a short diagnostic that preserves facts and checks for correctable errors.
A three-branch check
- Was the application submitted? Find the confirmation, application ID, timestamp, or portal status. If none exists, check the portal and deadline before assuming a screening decision occurred.
- Is there an observable mismatch? Compare required fields and any preview with the resume and answers you submitted. Correct a specific error if the application is still editable.
- Is the concern about access or accommodation? Save the exact step and error, then use the employer’s published accessibility or accommodation channel. For U.S. disability questions, consult the EEOC guidance linked below; rules elsewhere differ.
If all fields are accurate and the only evidence is a generic rejection notice, record the cause as unknown. That distinction keeps the troubleshooting useful without turning a suspicion into a claim.
Worked example: from rejection to a documented next step
- Task: A fictional applicant wants to understand an unsuccessful application for a customer-support coordinator role.
- First output/attempt: “The AI rejected me because I did not repeat enough keywords.”
- Mistake or weakness: The rejection notice says only that another applicant was selected. It does not mention AI, keywords, or an error, so the claim assigns a cause the applicant cannot observe.
- Human correction: The applicant saves the posting, submitted resume, confirmation, and notice; checks the portal for missing fields; and compares the work-history preview with the source file. The preview is accurate and the application is closed. The candidate does not create another account or resend the application.
- Quality check: Mark evidence as “confirmed” only if it appears in the saved records. In this illustrative log, the receipt and submitted file are confirmed, the portal shows no field mismatch, and the reason for selection remains unknown. The two confirmed items are not proof of automated screening.
- Final usable result: The applicant writes: “Submitted 12 October; receipt saved; no portal error visible; rejection notice gives no reason; screening method unknown.” They send one neutral question through the recruiter contact printed on the posting, then preserve the answer—or the absence of one—without changing their resume to chase an imagined score.
What to ask—and what to do when the application is closed
When a real field is wrong, ask about that field: “I applied for [role] on [date]. The preview showed [field] as [incorrect value], although the file I submitted lists [correct value]. Can the application record still be updated?” Use only true details and the employer’s verified contact route.
For the example above, there is no mismatch to correct. A reasonable optional question is: “If feedback is available, could you share whether I missed a stated qualification?” If the employer says feedback is unavailable—or does not answer—mark the case “closed; reason unknown” and continue. Do not claim the message proves AI use, and do not keep sending follow-ups to a closed requisition.
Make a small evidence record
| Item | Example entry | What it establishes |
|---|---|---|
| Notice | “Another applicant was selected” | Only the notice wording; not the screening method. |
| Submission | Receipt and saved PDF present | Application submitted with that file. |
| Portal fields | No visible mismatch | No observable import error in the preview checked. |
| Cause | Unknown | No claim about AI or keywords is supported. |
| Next action | One neutral question; move on if unavailable | Concrete follow-up without asserting an unseen cause. |
Beginner exercise: audit one application without guessing
Use one saved application. Gather the posting, submitted file, confirmation, portal fields or preview, and rejection text. Create four columns—fact, source, what it establishes, next check. Put “AI used” in the hypothesis column unless the employer confirms it. If a specific accessibility or disability-related issue appears, record it and use the published contact route. Keep the completed note private and remove customer or employer-confidential data.
Pass check: Keep the posting, submitted file, dated portal notice, any visible imported fields, and a copy of one neutral inquiry if you send it. If the employer replies, record only what it confirms. If there is no response, stop after the stated timeline and move on; do not fill unknowns with an AI-rejection theory.
Frequently asked questions
Can an applicant tell from a rejection whether AI was used?
Not from a generic notice alone. A message may not identify the tool, selection stage, or reason for the decision.
Should I add more keywords after an unexplained rejection?
Only use accurate terms that describe work you have actually done. One unexplained outcome is not evidence that a keyword caused it.
Can I ask the employer for an explanation?
If a recruiter or feedback channel is available, you can ask once, neutrally. The employer may not provide individualized feedback or disclose internal tools and scores.
A relevant next step
Use this only if the record shows a document or application-field issue. the ATS-friendly file guide. Use this to check terms when the posting itself identifies a missing requirement. the keyword-selection guide.
Compare the record, the notice, and your next action
Keep the rejection guide narrow: the question is what can be verified after an unsuccessful application, not how to optimize a resume for an imagined model. First confirm the role was open when you applied and that the portal recorded a completed submission. Next, preserve the exact resume and application answers. Then classify the result as one of three things: an observable technical or accessibility problem, a stated qualification or eligibility mismatch, or an outcome whose reason is not provided.
| Evidence | What it supports | What it does not support |
|---|---|---|
| Confirmation number and submitted-file copy | That an application was recorded with this file. | That a human or AI reviewed it, or why the result occurred. |
| Portal field shows a different end date | A visible discrepancy to correct or report. | That this discrepancy caused a rejection. |
| Notice cites an unmet license requirement | A stated qualification reason in that notice. | Any broader claim about automated screening or unlawful bias. |
| Generic decline with no explanation | The application was declined, if the portal confirms it. | The decision method or specific cause. |
One concise inquiry is usually more useful than repeated messages. Use the recruiter or support address named by the employer, include the role and submission date, and ask about one observable issue. If the employer cannot provide individual feedback, mark “no response available” and return to the search. Do not contact multiple employees with the same demand or treat a lack of reply as evidence about a tool.
Match the next step to what the record shows
If a visible field is wrong: use the employer’s correction channel and provide the accurate value. If the notice says the role is closed or filled: save it and do not treat a generic status as evidence of an automated decision. If the rejection is unexplained: you may ask a brief, neutral question about the application process, but the employer may not disclose internal tools or scoring.
Keep a separate branch for disability or discrimination concerns. The EEOC’s ADA-and-AI worker sheet explains that reasonable accommodation may be relevant to some technology-based assessments and points to additional resources. That guidance does not prove that AI was used in your case or guarantee a reversal. Preserve dates and notices, ask for an accommodation when appropriate, and seek qualified advice if you believe a protected right was violated. Do not send sensitive medical details in a general status inquiry.
Sources
- U.S. EEOC: Tips for Workers—The ADA and AI — Gives examples of employers using technology to score applications or tests and points to disability-related rights and resources; it does not identify why an individual was rejected. Checked 2026-10-17.
- Greenhouse Support: Unsuccessful resume parse — Supports the possibility of a parser issue in Greenhouse specifically, not a general conclusion about other ATS products. Checked 2026-10-17.