Your PDF résumé: what a hiring system reads
You send the same file to dozens of roles and never see what happens on the other side. We see it, because we read your résumé before filling in any form, and because we got it badly wrong doing so.
What happens to your PDF before anyone reads it
Almost every job site does what we do: open the file, pull the text out, store the text. The good-looking résumé you built becomes a strip of words, and it is that strip which fills fields and reaches a recruiter. Extraction hands the text back in the order the file stores it, not the order you see.
Does a two-column PDF survive?
Nobody publishes what a given hiring system does inside. What we can say is what happens on our side, with ordinary tools. Two columns are not forbidden, they are a risk: if the file stores the boxes side by side, a line from the left comes out stuck to one from the right. The same goes for your name inside an image.
There is a ten-minute test that settles this without theory. Open your PDF, select everything, copy it and paste it into a plain text editor: what shows up there is roughly what the programs see. If it came out scrambled, export it again with a simpler layout.
The six things that break most
| Two columns side by side | lines from different columns mix |
| Name or contact inside an image | vanishes entirely, it is not text |
| A scanned PDF with no text layer | returns zero characters |
| The generator's header and footer | come out of order, or vanish |
| A link written as a word | the address hides inside the file |
| An accent stored in two pieces | turns into PRODUÇA O mid-sentence |
The link case deserves a line. The contact line is written like this: email, LinkedIn, Blog. The address lives hidden in the link target, so a reader that takes only the words keeps the word LinkedIn where an address should be. We read the targets too now.
Why does my name come out wrong on the form?
Because the name is read by shape, and shape lies. The rule any reader uses: near the top, a line of two to five capitalised words with no numbers. A section heading looks exactly the same.
We measured it in our own database on 30 August 2026. One résumé had the contact line above the name, so the first line that looked like a name was the heading Technical Skills. That became the person's name, and two applications reached real employers signed that way. Extraction now skips known headings, and with nothing that looks like a name the field is left empty.
A résumé that survives any reader
- One column. If you like two, run the copy-and-paste test before you send it.
- Name on the first line, alone. Job title under it, contact under both, never above.
- Real text, never an image. Name, email and phone have to be selectable with the cursor.
- Write the address out. Put linkedin.com/in/yourname instead of hiding the address behind the word LinkedIn.
- Common section headings and dates in numbers. Experience, Education, Skills, Languages; Mar/2021 to Aug/2024.
A field the document does not support is left empty, and what is empty becomes a question for you, never a claim to an employer. The résumé check shows you that reading on your own file, and auto-apply shows where it sits in the cycle.
Next to this one, the Greenhouse form and how a robot that applies works. Prices on Plans and pricing, the rest on the FAQ. Create an account and upload your résumé.