How SmartRecruiters Reads Your Resume
SmartRecruiters is one of the few applicant tracking systems whose parsing engine is a matter of public record: its chief executive named Textkernel in an interview. That is unusually useful, because Textkernel publishes detailed developer documentation about what its parser does and does not handle — far more than SmartRecruiters says itself.
What SmartRecruiters accepts
Every line below links to the page it came from. Where SmartRecruiters does not publish a detail, this guide leaves it out rather than guessing.
Textkernel — Named publicly by SmartRecruiters' chief executive, who described roughly 98% parsing accuracy across application types after adopting it.
What ATSPrep exports, and whether SmartRecruiters takes it
This is computed at build time by comparing our export formats to SmartRecruiters's published list — not written by hand, so it cannot drift out of date.
| ATSPrep export | Accepted by SmartRecruiters |
|---|---|
| Yes | |
| DOCX | Yes |
What trips people up here
2 MB is a tight ceiling
SmartRecruiters' application endpoint documents a limit of 2 MB per file — among the lowest of any major system. A resume with an embedded photograph or a design-tool export will clear that limit easily, and a plain text-based export will not come close to it.
Source: SmartRecruiters Developers — posting an applicationThe parser's own authors are wary of PDF
Textkernel's documentation says that to minimise conversion problems you should not use PDF documents, on the grounds that most of their bad-parse reports come from corrupt PDFs, and that PDF often hides issues with the underlying text. That is a warning about how the file was produced rather than about the format itself: a PDF generated as clean linear text is fine, and the only way to know which one you have is to extract the text back out and look.
Source: Textkernel Developers — resume parser, getting startedDesign-led resumes are a named failure mode
Textkernel calls out artists and graphic designers specifically — resumes using images instead of text, and diagonal or vertical text — along with the very long CVs common in academia and medicine. If your field encourages a visually distinctive resume, this is the engine most likely to punish it.
Source: Textkernel Developers — resume parser, getting startedImages are accepted, which is not the same as read
JPG and PNG appear on the accepted list. Textkernel supports image formats only with an optional OCR add-on, so an image resume may be stored without its text ever being recovered. Accepting a file and understanding it are different things.
Source: Textkernel Developers — supported file formatsFormatting for SmartRecruiters
Stay well under 2 MB
This is the tightest limit among the major systems. Strip embedded images and avoid design-tool exports, which carry large font and graphic payloads for no benefit to a reader.
Send text, never a picture of text
The upload will take a JPG or PNG. The parser needs an OCR add-on to make anything of it, and you have no way of knowing whether the employer pays for one.
Verify the PDF rather than avoiding it
Textkernel's objection is to PDFs that hide broken text, not to the format. Export the file, extract the text back out, and confirm it reads in order before you send it.
Keep the layout linear
Diagonal text, vertical text and image-based headings are named directly. Whatever the visual convention in your field, the parsed record is what gets searched.
Read your resume back the way SmartRecruiters will.
Guidance is worth less than evidence. Open the studio, export a PDF, and the X-Ray re-opens that exact file and shows you the text a parser recovers from it — in reading order, with anything it dropped marked. Free, no account, and your file is never uploaded.
Open the Studio →SmartRecruiters and your resume
What file types does SmartRecruiters accept?
PDF, DOC, DOCX, RTF, JPG and PNG, with a documented limit of 2 MB per file on its application endpoint. Note that accepting an image format is not the same as reading it — extracting text from an image requires OCR, which is an optional add-on to the underlying parser.
What parser does SmartRecruiters use?
Textkernel. SmartRecruiters' chief executive named it publicly, describing about 98% parsing accuracy across application types and citing the need for a partner that could scale with rapid traffic growth and support multiple languages.
Should I avoid PDF for SmartRecruiters?
Textkernel's own documentation advises against PDF to minimise conversion problems, saying most of its bad-parse reports come from corrupt PDFs and that the format often hides problems with the underlying text. The practical reading is narrower than the headline: a PDF built as clean linear text parses fine. What matters is checking the text can be extracted rather than trusting how the page looks.
How large can a resume be in SmartRecruiters?
2 MB per file according to its developer documentation, which is one of the lowest ceilings among the major applicant tracking systems. A plain text-based resume sits comfortably below it.
How the rest of them read a resume
Greenhouse
Greenhouse takes PDF, DOC, DOCX, RTF and TXT, and will not parse a resume larger than 2.5 MB. Its own support docs name what makes a parse fail — columns, contact details in a header, abbreviated job titles and more.
Oracle Taleo
Oracle publishes Taleo's accepted attachment formats but ships no resume parser of its own — parsing only works if the employer has enabled a third-party partner integration, so behaviour differs from one Taleo site to the next.
Ashby
Ashby accepts PDF, DOC and DOCX up to 50 MB but does not support Markdown. Its docs name the two reasons a resume fails to parse, and its better parser is an opt-in feature some employers never switch on.
Workable
Workable accepts PDF, DOC, DOCX, RTF, HTML and ODT for a resume, with a 5 MB ceiling — while other attachments on the same form are allowed four times that size.