Multi-column CVs break because text extraction reads a page as a stream, not as a picture. A two-column layout can be read across the page rather than down each column, interleaving your job titles with unrelated skills. The result is a document that looks polished to you and arrives as scrambled fragments. One column, no text boxes, no tables for layout.
What Extraction Actually Does
When a system reads your CV it pulls out a sequence of characters and then tries to work out what each part means. Position on the page is a weak signal at best, and in many extraction paths it is discarded entirely.
So a layout that depends on a human eye understanding "this column is a sidebar" has no equivalent instruction for a parser. The sidebar text simply appears somewhere in the stream, often mid-sentence through your employment history.
The Classic Failure
A left sidebar containing skills and contact details, with experience on the right. Read line by line across the page, the output becomes "Python Senior Analyst SQL March 2023 Excel Barclays" and so on. Every word is present. None of it means anything.
Why Design Templates Cause This
Template tools optimise for how a document looks, which is the correct priority for a poster and the wrong one for a CV. The attractive layouts are attractive precisely because they use columns, panels, icons and shaded blocks, and each of those is a structure that carries no text meaning.
The worst case is a CV exported as a flat image inside a PDF wrapper. It looks immaculate. It contains no extractable text at all, so an automated read returns an empty document and a human may never see it.
Elements That Cause Trouble
- Two or three column layouts, including narrow sidebars
- Text boxes, which frequently extract out of order or not at all
- Tables used to position content rather than to present data
- Headers and footers, which are commonly dropped
- Icons replacing words, so a phone symbol conveys nothing
- Shaded panels and graphics containing text
- Skill bars, star ratings and charts
The Test That Settles It
Open your CV, select all, copy, and paste into a plain text editor such as Notepad or TextEdit in plain text mode. What appears is approximately what a parser receives.
- Is your name near the top, followed by your contact details
- Do job titles, employers and dates stay together
- Does each bullet remain one coherent sentence
- Has anything vanished entirely
- Has anything arrived in an order you did not intend
If the pasted text reads sensibly, the layout is safe. If it does not, no amount of rewriting the content will help, because the content is not reaching the reader.
A Layout That Always Survives
Single column, full width. Name and contact details as plain text on the first two lines, in the body rather than a header. Section headings as real headings. Roles as title, employer, location, dates on one line, then bullets beneath.
That structure is not dull by necessity. Typography, spacing and a restrained use of one accent colour produce a document that looks considered and still extracts perfectly. What you give up is the sidebar, and the sidebar was never carrying its weight.
Making Single Column Look Good
- Generous white space between sections rather than boxes to separate them
- One clear typeface at 10 to 12 point, with bold for emphasis
- Consistent alignment of every date, on the same edge throughout
- Section headings slightly larger and bold rather than reversed out of a colour block
Where Design CVs Are Genuinely Expected
Graphic design, architecture and some creative roles do expect visual craft, and a plain document can undersell you. The workable answer is two documents: a designed portfolio piece for human review, and a plain single-column version for the application system.
Send the plain version through the portal and link the designed one. Where an advert explicitly invites a portfolio, follow that instruction, but do not rely on a visual CV to carry structured information through an automated stage.
Headers, Footers and the Lost Phone Number
This is the most damaging single error and the least visible. Contact details placed in a document header often do not appear in extracted text, producing a CV that reads well and identifies nobody.
Page numbers in a footer are harmless. Your name, phone number and email must sit in the body of the document.
Fixing a Template You Have Already Paid For
Plenty of people have bought a template and would rather adapt it than start again. Most can be salvaged, because the damage usually comes from three features that can be removed without rebuilding.
The Three Removals
- Collapse the sidebar into the main column, keeping the content and losing the panel
- Replace every text box with ordinary paragraphs in the body
- Move anything in a header or footer into the first two lines of the document
What survives is the typography, the spacing and the colour, which is most of what made the template attractive. What goes is the structure that broke extraction.
Run the copy and paste test after each removal rather than at the end. That way you learn which single feature was causing the scrambling, which is worth knowing for the next document you build.
Frequently Asked Questions
Will a Two-Column CV Definitely Be Rejected?
Not definitely. Some extraction handles columns correctly and some human readers will see it as intended. The point is that it introduces a risk with no compensating benefit.
Are Tables Always a Problem?
Tables presenting genuine tabular data, such as a list of qualifications with dates, usually extract acceptably. Tables used invisibly to position blocks of content are the ones that scramble.
What About PDF Versus Word?
Either is fine if the PDF contains real text rather than an image. Export from a word processor rather than a design tool, and run the copy and paste test afterwards.
Can I Use Icons for Contact Details?
Use the word as well. An envelope icon next to an address is fine; an envelope icon instead of the word "Email" leaves nothing for a parser to label the field with.
Do Colours Cause Problems?
Colour itself is harmless. Text rendered inside a coloured graphic is the problem, as is light grey text at low contrast, which is hard for people to read even when it extracts.
A Fifteen-Minute Rebuild
Copy your content into a blank single-column document, put contact details in the first two lines of the body, use plain headings, and delete every box, icon and rating bar. Then run the copy and paste test. Most people find the plain version reads better on screen as well, because the content was doing the work all along.
InsightCircle
