Career Advice for Job Seekers
Paste Your Resume Into Notepad. That’s Roughly What the Software Got.
By Jack Allen, Founder, Passthrough
Here is a resume that looks fine on screen: clear headings, readable fonts, contact details in a tidy left column, work experience on the right. This is what a text extractor made of it:

The resume is fictional. The output is real: a two-column layout built as a Word table, exported to PDF and read with the free tool pdftotext.
The email address has landed under the Experience heading. The phone number turns up after the last bullet of the second job. Nobody rejected this resume. It was misread, and the misreading happened before a single person looked at it.
That gap is the subject here, and you can check your own resume for it in about a minute.
Your resume has two readers
Students are often told that an applicant tracking system, or ATS, is a robot gatekeeper that deletes resumes. The documentation tells a narrower story. When you submit a file, the system turns it into text, stores it, and lets recruiters search and filter the people who applied. So your resume has two readers: software that extracts the text, and a person who searches and skims what the software extracted.
Greenhouse, one widely used system, describes automatic rejection as a rule the employer sets on your answers to application questions. Its own blog uses a Class A commercial driver’s license question as the example: applicants without the license can be rejected automatically for a tractor-trailer job. That is the one place a careless click can end an application with no person involved. Read every required question slowly and answer it truthfully.
The rest of the process is quieter. Greenhouse lets recruiters search resumes for keywords, and its guidance says a keyword must match the application exactly to appear in the results. A search can only find text that survived the trip from your file into the system.
Other systems behave differently, and nobody outside an employer knows exactly how theirs is set up, me included. What I can show you is the step every one of them has to do first: turn your file into text.
That step can fail. Greenhouse’s help page for recruiters who upload a resume to fill in a candidate profile lists what can break the import: resumes saved as images, complex tables, headers and footers, a name or contact details in a header, footer or text box, and files over 2.5 MB. When it fails, the profile fields don’t fill in on their own. That page is about recruiters importing a file, not a promise about every application, but it points at the same place my own test did.
So the realistic risk is not a robot that rejects you. It is a resume that arrives scrambled, half-filled or blank, in front of someone with a pile to get through.
The Notepad Test
It takes 60 seconds and needs no tool, account or download.
- Open the exact file you plan to upload: the PDF, or the Word file, not the draft you made it from.
- Select all and copy (Ctrl+A then Ctrl+C on Windows, Cmd+A then Cmd+C on a Mac). If nothing highlights, your file is a picture of text, and software will see a blank page.
- Paste into Notepad on Windows, or TextEdit on a Mac after choosing Format, then Make Plain Text.
- Read it top to bottom the way software would.
Then ask five questions. Is your name first? Are your email and phone number complete and near the top? Is every job still under its own heading, in order? Does any line mix words from two different places on the page? Is anything missing?
Now, a paste shows how your own program read the file, which is one reader among many. A messy paste is a warning. A clean paste means you cleared the cheapest hurdle, not every hurdle.
What I tested
I built one fictional resume and saved it eleven ways, using layouts students really use. Ten were made in Word and exported to PDF. The eleventh is a PDF I built with a script, so I could control the order its text is written in. Each file went through three text readers: two free, open-source tools for its file type, and a parser I built myself. The files came from LibreOffice and a script, so a resume exported from Word, Google Docs or Canva may behave differently.
Here is what happened.
- Single column. Every reader agreed. This is the baseline.
- Two columns, built with Word’s column setting. All six reads kept everything, in order. Greenhouse lists columned layouts as a risk, so this surprised me. It doesn’t make columns safe everywhere. It means I could not break this one.
- Two columns, built with a table. Five of six reads kept everything in order. The sixth, a Word-to-text converter, laid the columns side by side, so every line mixed both.
- Skill-level dots (●●●○○). Read as ordinary text. Harmless.
- A sidebar in a text box. Six reads of the same sidebar gave four different results. One reader put the sidebar before the name. Two put the contact details after every job. One dropped the sidebar entirely: contact details, education and skills, gone. Two read it fine.
- Contact details in the page header of a Word file. Two readers dropped them completely. The others kept them.
- Contact details as a picture. All six reads lost them. A picture of your email address is not an email address.
- Sections side by side in a table, row by row. Every word survived, but sections got mixed. In four of six reads, the second job ended up under the Education heading. In another, both jobs came between the email address and the phone number.
- Headings with extra space between the letters. This is a common design touch. The Word readers handled it. All three PDF readers returned “E X P E R I E N C E,” and my parser then could not find the Experience, Education or Skills sections in the PDF. Greenhouse’s list names spaced-out letters too.
- A PDF that stores its two columns row by row. I built this one by script, so treat it as a demonstration of the mechanism. I did not test any design tool. Three readers gave three results. One mostly straightened it out but glued a job title onto the education line. One mixed the contact details into the experience. One read each row straight across, so “CONTACT EXPERIENCE” came out as one line.
- A PDF that is really a picture of a page. Blank. Nothing to read.
Three of the eleven layouts were read cleanly by every reader. In the other eight, at least one reader stumbled, and none of the eight looks wrong on screen. Even one free tool disagreed with itself: for three of the eleven PDFs, pdftotext returned the words in a different order depending on its mode. If one tool can’t agree with itself, you can’t guess what another program will do from how the page looks.
What to do with this
Looks were rarely the problem on their own. What broke was words that never reached the software, arrived in the wrong order, or came out with spaces between the letters. That points to a short list.
- Put your contact details in the body of the page, as plain text, right under your name. Not in a header, a text box or a picture.
- Keep each section together. If you use columns, keep a whole section in one place, and run the test.
- Avoid text boxes for anything you need read.
- Use normal spacing in headings. Skip letter-spaced titles like E X P E R I E N C E.
- Export a PDF from Word or Google Docs, not a screenshot. If you cannot select the text, software cannot either.
- Use plain headings such as Experience, Education and Skills, so every kind of reader can tell where a section starts.
- Write job titles out in full. Senior, not Sr.
- Keep the file small. Greenhouse says it can’t parse resumes over 2.5 MB, and big files usually mean images.
Words, not tricks. People who search resumes use the words in the job posting, and a search wants an exact match. If you did the work the posting describes, say so in the posting’s plain terms. Do not hide extra keywords in white text. Anything in the file shows up in a plain-text paste, and the person reading it will see it too.
What this test can’t tell you
It doesn’t tell you how any employer ranks, scores or filters. It checks the one part you control: whether your words come out in one piece. I looked at one vendor’s documentation and tested eleven layouts from one fictional resume, so treat the results as a pattern worth checking on your own file, not a law.
And a clean paste is only the start. After that, a person still reads your resume for evidence that you can do the job.
A person can squint at a messy layout. Software can’t. Paste your resume into Notepad and find out which one you’ve been designing for.
For career advisors: a five-minute exercise. Have each student run the Notepad Test on their own resume, then swap screens with a neighbor. The neighbor has ten seconds to answer three questions: who is this, what have they done, and how would I reach them? Ask each pair for one thing that moved, vanished or ended up under the wrong heading, and for the fix. No account, no upload, nothing leaves the room.
— Jack Allen is the founder of Passthrough, a free tool that checks a resume against a specific job posting. Its Robot’s-Eye View shows the text a parser pulled from your file, with no account needed, and nothing is stored.
New Job Postings
Advanced Search