ATS screening explained: parsing is not selection
An applicant tracking system helps an employer manage applications. “ATS screening” can refer to different steps. Understanding which one you mean is more useful than trying to beat an imaginary universal algorithm.
1. Parsing: turning a document into fields
Parsing extracts information from a CV or resume into a candidate profile. A document can look clear while some text is difficult for a parser to interpret.
Greenhouse describes extraction that fails or is partial because of file size or formatting. Its guidance says a failed parse leaves the resume attached and requires manual input of details. This is not the same as automatic rejection.
Follow file instructions, use clear headings and inspect populated fields. A plain-text copy check may expose reading-order problems, but cannot certify compatibility with the employer’s system.
2. Screening questions: checking specific requirements
A portal may ask about qualifications, availability or eligibility. These are separate from the words in your CV. In Oracle Recruiting, disqualification questions can determine whether a candidate moves forward or is automatically disqualified, based on configured requirements and answers.
This is one system’s documented capability, not a claim about every employer. Answer honestly. Adding keywords cannot change whether you hold a required qualification.
3. Matching: interpreting the evidence you provide
A job-specific CV check compares your information with an advert. It can help identify a requirement you have not clearly evidenced. It cannot know the employer’s full configuration, internal priorities, other candidates or final decision.
A third-party match score is that tool’s interpretation of its inputs, not a score issued by the employer’s ATS. The vendor sources cited here establish no universal pass score. A higher tool score is not a promise of an interview.
What CareerKitHub can and cannot help with
Use the check on this page to compare CV text with an advert and decide where evidence needs clarification. Treat suggested wording as a draft to verify.
Do not treat it as certification that your original PDF parses correctly, proof of legal or professional eligibility, or an explanation of a previous rejection. Rules-based checks follow fixed logic; language models can misinterpret or invent details. Neither has access to the employer’s private process.
A sensible pre-application routine
- Read the requirements and application instructions.
- Make your actual relevant experience easy to find.
- Inspect the exported document and application fields.
- Review screening answers separately.
- Submit only claims you can support.
Use the ATS checklist alongside this routine. Choosing a workflow? The CareerKit alternatives guide compares tasks rather than promising better hiring outcomes.
Check one CV against one job
Have your CV or resume and one job advert ready. Paste them into the check on this page. It calls the same guest check as Studio, and only after you tick consent. Exploring rather than applying? Use the fictional Amelia Vance sample. That sample stays in this browser until you tick consent and run the check.
Free CV check on this page
Paste one CV or resume and one job advert. The button calls the existing free check. The first check in this browser does not ask for an email. A later day asks for an email, or for you to sign in, before another free check. It does not open payment.
This is guidance, not an employer’s ATS result. When the review service is configured, the check can return the requirements it compared and the next steps it suggests. A rewritten line is shown only when the response includes one. If the service is not configured, or it cannot finish a verified reading, the same endpoint can return a rules-based reading on CareerKitHub’s server. A rules-based reading cannot certify that an employer will parse an original file. A model reading can misunderstand context or suggest unsupported wording. Review every suggestion. Never add a skill, qualification or result you cannot substantiate. No score guarantees an ATS pass, an interview or a job. A free check does not mean later rewriting or downloads are free.
You can read the result without an account. Signing in is how a review is saved in Studio. This page does not upload a file and does not take payment.
Before you paste
Remove unnecessary contact details and confidential information. This page accepts pasted text only. The first check in this browser does not ask for an email. If this browser already completed a free check on an earlier day, a new day asks for an email and a separate tick, or for you to sign in. That email is a lead record described in the privacy information. This page does not create an account from the check, and it does not save the CV, the job advert or the result to an account. The text stays in this browser until you change it or leave the page.
The consent box starts unticked. Nothing is sent until you tick it and run the check. The privacy information covers this guest check. Anthropic receives the text only when the review service is configured. Otherwise, or if a verified reading cannot be finished, CareerKitHub can return a rules-based reading on its own server. The daily caps can be configured: one for a connection that has an address, one shared by guest checks, and one used when no address is available. A recorded count stores a hash and a count. It does not store the CV text. Provider retention and backups still apply to text that reaches Anthropic. This page does not promise that a provider deletes that text.
Creating an account and uploading a CV is a later step. The file is processed in memory and not stored, and the extracted text, the job advert and the results are saved to that account. In Studio, a PDF, DOCX or TXT can be read in memory; the original file is not stored.