# How to write a CV that screening software can read Updated October 2026 Screening software reads a CV in 4 steps. It extracts the text, splits it into fields such as job titles and skills, compares them with the job advert and ranks the applications for a recruiter. A CV gets through when its text comes out clean and uses the words in the advert. ## What the software does with your CV Larger employers usually run applications through an applicant tracking system, or ATS. It does four things with your file: 1. It extracts the text. Anything that isn't real text, such as a scanned page or words in an image, comes out empty. 2. It splits the text into fields: name, contact details, job titles, employers, dates, education and skills. It finds them by their headings and their position on the page. 3. It compares those fields with the advert: the job title, the required skills, the tools and the years of experience. 4. It sorts or filters the applications for a recruiter, who reads the parsed version first. The recruiter sees what the parser made of your CV, so a strong CV can still read badly. ## A layout that survives parsing - Use one column. Two columns often get read across the page, which mixes your job titles with your skills. - Use the headings people expect: Experience, Education, Skills. A heading like "My journey" leaves the parser guessing. - Put your name, email and phone number in the body. Many parsers skip page headers and footers. - Keep text out of images, icons, charts and text boxes. Skill bars and star ratings carry no text a parser can read. - Write dates the same way every time, such as "March 2022 to June 2024". - Send a Word file or a PDF exported from a word processor. A scan or a PDF saved as an image has no text in it. ## Words that match the advert The match score counts the words in the advert, so use them where they describe your work honestly. - Use the advert's job title for your current role if it's the same job under a different name. - Name the tools you use. The match score looks for the product names an advert lists, and our data holds 8,753 of them across occupations. - Write an acronym out in full once, then use the short form: "know your customer (KYC) checks". - Put the skills the advert lists in a Skills section and in the bullets where you used them. Don't hide keywords in white text. The recruiter reads the parsed text, so hidden words show up there. ## Lead with the work AI can't do Across the 402 jobs we've graded, an average of 38% of task time still needs a person. That share is mostly live conversations, hands-on work and decisions someone signs off. Put that work near the top of each role, with a number attached, such as the size of the team, the value of the decisions or the number of cases you handled. The [skills that keep work human](https://www.aigentcy.com/ai-exposure/skills/) page lists them by job, and each [occupation page](https://www.aigentcy.com/ai-exposure/) shows which of its tasks AI does least well. ## Check your CV before you apply [The CV checker](https://www.aigentcy.com/tools/cv-checker/) shows what screening software reads from your file and how it matches an advert. [The AI exposure checker](https://www.aigentcy.com/tools/ai-exposure-checker/) reads the tasks in your CV and scores how much of your work AI can do today. ## Questions about CV screening Does screening software reject CVs automatically? It depends on how the employer sets it up. Knockout questions, such as right to work or a required licence, can screen an application out straight away. Keyword scores usually sort the pile for a recruiter to read. Should I send a PDF or a Word file? Both work when the file holds real text. A scanned PDF has none. If the advert asks for a format, use that one. How long should my CV be? Length matters more to the recruiter than to the software. Two pages covers most experienced roles. One page is enough early in a career. Will AI screening read my CV differently? Newer systems use language models, which recognise synonyms and related job titles. They still read the extracted text, so the layout rules above still apply. ## Related topics - [**Jobs AI can't replace**Cement masons score lowest, at 8.](https://www.aigentcy.com/ai-exposure/jobs-ai-cant-replace/) - [**Jobs most at risk from AI**Proofreaders score highest, at 76.](https://www.aigentcy.com/ai-exposure/jobs-most-exposed-to-ai/) - [**High-paying jobs with low AI exposure**Radiologists earn a median $420,860 and score 46.](https://www.aigentcy.com/ai-exposure/high-paying-jobs-low-ai-exposure/) - [**Entry-level jobs with low AI exposure**Cement masons need little prior training and score 8.](https://www.aigentcy.com/ai-exposure/entry-level-jobs-low-ai-exposure/) - [**Jobs AI speeds up the most**For natural sciences managers, AI speeds up 96% of task time.](https://www.aigentcy.com/ai-exposure/jobs-ai-speeds-up/) - [**AI and jobs in Malta**15 iGaming roles graded task by task.](https://www.aigentcy.com/ai-exposure/malta/) - [**Skills that keep work human**The skills to build in the jobs AI changes least.](https://www.aigentcy.com/ai-exposure/skills/) - [**Jobs A to Z**Every job we publish, by first letter.](https://www.aigentcy.com/ai-exposure/jobs/) - [**AI exposure by industry**Average scores for 22 occupation groups.](https://www.aigentcy.com/ai-exposure/industries/) - [**AI for businesses**The tasks to automate first in each team and industry.](https://www.aigentcy.com/ai-for/) - [**How we score AI exposure**The grading, the weights and every source.](https://www.aigentcy.com/ai-exposure/methodology/) ## Next steps - ### Check your job Your job title is a rough guide. Your actual tasks decide your score. Paste your job description or CV and see which of your tasks AI can do. [Check your job](https://www.aigentcy.com/tools/ai-exposure-checker/) [Check your CV](https://www.aigentcy.com/tools/cv-checker/) ### Check which roles in your team AI can take on We map your team's tasks, show what AI can do today and build the automations that pay back first. [Book an AI audit](https://www.aigentcy.com/audit/) Occupation data from the O\*NET 30.3 Database by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA), used under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) and modified by Aigentcy. O\*NET® is a trademark of USDOL/ETA. Wage, employment and projection data from the U.S. Bureau of Labor Statistics. EU occupation names from ESCO, European Commission. AI usage signals from the Anthropic Economic Index (CC BY) and Microsoft Research (CC BY 4.0). [Methodology and sources](https://www.aigentcy.com/ai-exposure/methodology/). Scores © Aigentcy. Quote them with a link to the page.