Isolve

Nov 16, 2026

How ATS software reads a resume, including Workday, Greenhouse, Lever, iCIMS, and Taleo

An ATS reads a resume in two steps. First it stores the file. Then it tries to parse it. Parsing means pulling out words and guessing which words are your name, your jobs, your schools, and your skills.

The stored file and the parsed text are not always the same. A recruiter might open the original PDF and see a clean page. The search box might be looking at a scrambled text version. That gap is why a resume can "look fine" and still be hard to find.

The systems people ask about

These are real products. Companies buy them. They are not magic, and they are not all the same.

Workday is used by many large employers. Applications often go through a Workday form. The resume file is still uploaded, and Workday also asks you to fill in fields. If the file is a picture, the form fields may stay empty and you have to type them.

Greenhouse is common at many tech companies. It stores the application and gives the hiring team a page with the resume. Parsing still matters when someone searches the database later.

Lever is another hiring system used by growing companies. It keeps resumes and notes on a candidate. Search still depends on text the system could read.

iCIMS is widely used by larger companies with lots of applicants. It has been around for years. Like the others, it can store a file and also try to structure the text.

Taleo is an older Oracle system that many big employers still use. Candidates often meet a long form plus a file upload. Text that is trapped in columns or text boxes is a known weak spot for older parsers.

None of these vendors publish a simple public rulebook that says "we reject resumes with two columns." Do not trust a blog that claims one secret setting. What you can check is your own file: if a plain-text extract is empty or mixed up, any of these systems may struggle.

Formatting that often breaks the text

Columns. Two columns look tidy. A parser may read across the page, so a job on the left gets mixed with a job on the right.

Tables. A skills table can come out as a pile of words in the wrong order. Plain lines are safer.

Text boxes and shapes. Words drawn inside a shape are often skipped. Put the words in the main story of the page.

Headers and footers. Your name and email in the header may never enter the body text. Repeat them at the top of the page.

Images. A logo or a photo does not add skills. If the whole page is an image, there may be no words at all.

Fonts. Unusual or script fonts can be substituted or dropped. A common font is boring and safer.

File type. A DOCX file is a zip of text, so many parsers read it well. A PDF can be fine if it contains real text. A PDF that is only a scan is not fine. An old .doc file is less reliable. Save it as DOCX or PDF first.

What to do with this

Open your file and try to copy the text into a blank note. If the paste is empty, out of order, or missing your email, fix the file before you send it. Use one column, real headings, and words you can defend.

Isolve's free preview tells you if text came out, and flags one or two clear problems. The full report shows the text itself, so you can see the order. It still cannot tell you which brand of ATS the company bought.