Most ATS advice circulating online was written for parsers that stopped existing years ago, and a fair chunk of it will actively hurt you. White text stuffed with keywords is a good example: it no longer sneaks you through, it gets your application binned and your name remembered for the wrong reason. What follows is a checklist of ten things that genuinely change how your CV is read, with the reason the software cares and what to do about it.
1. Give the parser one column to follow
Two-column CVs look great in a design portfolio and cause real problems in extraction. When text sits in side-by-side frames, some parsers read straight across the page instead of down each column, so your skills list ends up shuffled into your job titles and neither section makes sense.
Build the document as a single column running top to bottom. If you want visual separation, use white space, bold job titles and a thin horizontal rule. You lose nothing in readability and you remove the most common cause of scrambled output.
2. Use section headings the software has seen a thousand times
Parsers segment a CV by matching heading text against a dictionary of expected labels: work experience, education, skills, certifications, languages. That mapping is how your third job becomes a dated entry in the employer's database rather than a floating paragraph.
A heading like "Where I have made an impact" maps to nothing. Everything underneath it lands in a generic bucket or disappears. Keep the headings boring and put the personality in the bullets.
3. Write dates one way and never deviate
Tenure is calculated by subtracting a start date from an end date. Anything ambiguous breaks that arithmetic, and a role with an unparseable date range can be dropped from the structured record even though the text is still in the file.
Pick MM/YYYY and use it in every entry, including internships and short contracts. Write "Present" for your current role rather than leaving the field open. Mixing "Jan 2020" in one job and "01/2020" in the next is enough to confuse weaker parsers.
4. Keep the contact block in the body of page one
Headers and footers live in a separate layer of the document. Most modern systems read that layer now, but not all of them do, and there is no possible benefit to gambling your phone number on which one the employer bought.
Put your name, email, phone, city and LinkedIn URL as ordinary lines of text at the top of the first page. One detail people miss: a phone number split across a graphic and a text box often comes out as half a number.
5. Nothing load-bearing inside tables, text boxes or graphics
Skill ratings drawn as five-dot meters carry no information a machine can read. A skills matrix built as a table may be extracted column by column, producing a list that reads "Python Advanced SQL Intermediate" with no separation. Icons next to your email address are decoration, and decoration is where parsing errors hide.
The rule is simple: if a fact matters, it must exist somewhere as plain selectable text in the natural reading order. Keep the graphics if you like them, but never let them be the only place a fact lives.
6. Spell out every acronym once, with the abbreviation beside it
Recruiters search the applicant database with boolean queries, and those queries are literal. Someone hunting for SEO experience who types SEO will not match a CV that only says search engine optimisation, and the reverse is equally true.
Write it as "search engine optimisation (SEO)" the first time it appears, then use the short form. Do the same for certifications, methodologies and tool names that go by two names. This single habit probably wins more search hits than any layout change on this list.
7. Mirror the job ad's vocabulary, but only where it is true
If the posting says stakeholder management and your CV says client relations, you are describing the same work in a word the recruiter is not searching for. Read the ad, note the exact nouns it uses for the skills you actually have, and adopt that phrasing.
The honesty part is not a disclaimer, it is self-interest. Keywords get you onto the shortlist; the interview is where a fabricated skill collapses in the first five minutes. Postulit pulls your real LinkedIn history into a structured CV, which makes this kind of targeted rewording quick instead of a full rebuild each time.
8. Name the file like a professional would
Firstname-Lastname-Role.pdf. That is the whole rule. The filename appears in the recruiter's list view alongside forty others, and CV_final_v3(2).pdf is a free bad impression you are handing out for nothing.
Avoid spaces, accents and special characters in the name, since some upload pipelines mangle them. It has no effect on parsing quality, but it costs ten seconds and it is the only tip here a human notices directly.
9. Send the format they ask for, and stop panicking about PDF
The blanket rule "always send .docx" is out of date. Mainstream systems parse a well-structured, text-based PDF without difficulty, and a PDF protects your layout from reflowing on someone else's version of Word.
What genuinely breaks is a PDF exported as an image, a scanned document, or one produced by a design tool that stored the text as vector outlines. If the posting names a format, use that one and stop thinking about it. We have covered the PDF versus Word question and the readable file types in more depth elsewhere on this blog.
10. Tailor per posting and ignore the score
Third-party "ATS score" checkers grade your CV against their own scoring model, not against the employer's system. Run the same file through two of them and you will get two different numbers, neither of which the recruiter will ever see. They are useful as a proofreading pass and meaningless as a verdict.
What actually moves an application forward is fifteen minutes of tailoring: rewrite the summary line for the role, reorder your top three bullets so the most relevant one is first, and swap in the ad's vocabulary from tip seven. Do that per posting rather than sending one perfect generic CV to thirty companies.
The advice worth deleting from your notes
White text keyword stuffing was briefly clever around 2012. Today it is trivially detected, most systems surface the raw extracted text to the recruiter, and getting caught puts you on an internal do-not-contact list at a company you wanted to work for. The same goes for pasting the entire job description in one-point font at the bottom of the page.
Also retire the idea that the ATS makes hiring decisions. It stores, parses and searches. A human still reads the shortlist, and that human has opinions about a CV that reads like a keyword soup.
The five-minute self-test
Open your CV, select everything, and paste it into a genuinely plain text editor: Notepad, TextEdit in plain text mode, or an empty file in VS Code. What survives is close to what a parser extracts.
Read it top to bottom. Are the job titles still attached to the right employers? Did the dates come through? Is your phone number there? Did the skills section survive as a readable list? If the answer to any of those is no, fix the layout rather than rewriting the words, because the words were never the problem.