DEV Community: Anikó Juhász

Does an ATS Reject Your CV Before a Human Sees It? What the Data Actually Shows

Anikó Juhász — Thu, 09 Jul 2026 16:16:04 +0000

The advice sounds urgent: "Optimise your CV for ATS or it never gets seen."

Behind it sits a statistic — "75% of CVs are automatically rejected before a human reads them" — that has been repeated in blog posts, LinkedIn courses, and resume-writing services for years.

It's not true. The number was made up.

Where the 75% stat actually came from

Researchers tracing the origin of this figure found a single source: an unverified estimate from a company called Preptel, a resume service that no longer exists. No study. No peer review. No sample size.

The stat spread because it sounded plausible and because it created demand for exactly the service Preptel was selling.

That doesn't mean ATS doesn't matter. It means the fear is disproportionate — and the advice built around it is aimed at the wrong problem.

What ATS actually does

Applicant Tracking Systems are, at their core, databases. They receive applications, parse them into structured data, and help recruiters organise and search through that data.

They don't reject. They rank and sort.

A representative survey of recruiters across tech, healthcare, finance, and manufacturing found that 92% said their ATS does not automatically reject applications based on formatting, design, or structural issues. Only 8% of systems are configured for automatic rejection at all — and those trigger on specific conditions, not on font choices or column layouts.

An application with a two-column layout and a non-standard font doesn't get rejected. It may parse poorly, which causes problems further down the line — but it sits in the database alongside every other application.

What actually causes automatic rejection

Real automatic rejections come from knockout questions — the mandatory questions applicants answer before or just after uploading their CV.

These are binary. There's no grey area.

"Are you authorised to work in the United States?" → No → Rejected immediately.

"Do you hold a valid CPA licence?" → No → Done.

"Do you have at least 5 years of experience in this role?" → No → Out.

For developers in Central and Eastern Europe, this is the filter that ends most applications — not the CV. A developer in Poland applying to a US company's "remote" role gets automatically rejected because the listing was never actually open to them. Not because the keyword "Python" appeared too few times.

This is why checking whether a role is legally accessible to you before applying matters more than any formatting decision.

The timing problem

Even when a role is genuinely accessible, timing beats formatting.

52% of recruiters review applications in order received. Most roles fill their interview shortlist — 10 to 20 candidates — within 48 to 72 hours of the listing going live. Once that shortlist exists, the review stops.

A developer who applies three days after posting, with a perfectly optimised CV, gets no response. Not because of ATS. Because the recruiter already has their candidates.

The practical implication: applying early matters more than reformatting. A solid CV submitted within the first 48 hours beats a redesigned one submitted a week later.

What's technically true about ATS formatting

Not all the ATS advice is wrong. Some formatting choices do cause real problems — not rejection, but poor parsing that buries useful information.

Multi-column layouts. Two-column CVs built in Canva or Figma look good on screen. ATS parsers read left to right, row by row. A two-column document becomes a scrambled mix — skills from the left column and job descriptions from the right run together into unreadable lines of text.

Contact information in Word headers or footers. Some ATS platforms — particularly older Workday and Taleo configurations — ignore document headers entirely when parsing. A CV with the applicant's name and email in the Word header produces a well-structured profile in the database with no contact details attached.

Image-based PDFs. A PDF exported directly from a word processor parses at around 91% accuracy. A scanned document or image-saved PDF drops to around 4% — the system sees a blank page.

Non-standard fonts. Uncommon decorative fonts can break character encoding. Arial, Calibri, Times New Roman, Helvetica — these work reliably. Custom fonts often don't.

The easiest test: copy your entire CV into Notepad (plain text editor). If the sections stay in logical order, dates are readable, and nothing disappears — the document will parse correctly in any system.

What third-generation ATS changes

ATS technology has moved fast. The newest generation — platforms using large language models as their core parsing and evaluation engine — reaches around 97% extraction accuracy. These systems understand context, synonyms, and career trajectory in ways earlier systems couldn't.

"Amazon Web Services" and "AWS" are the same thing to a 2026 ATS. "Senior Engineer" and "Tech Lead" are evaluated in context, not as separate keywords.

The older advice — include both versions of every abbreviation, never use synonyms for your job title — still applies to companies running older systems (and plenty do). But for developers applying to companies using modern hiring software, the formatting anxiety is even less warranted than it used to be.

The real filters haven't changed: work authorisation, location requirements, minimum qualifications — and the job being actually open.

What this means in practice

Before you apply:

Check the posting date. Anything older than 45 days with no update deserves scrutiny. Ghost jobs — listings kept live with no real hire intent — account for roughly 40% of all job postings. No CV optimisation fixes a job that was never going to result in a hire.

Check the authorisation requirements. If the listing doesn't mention hiring outside the US or UK, or doesn't specify a contract type (B2B, EOR, or local entity), ask before you invest time in the application. The question takes two minutes.

Apply early. If the role looks right, submit within 48 hours.

Get the basics right. Single-column layout. Contact information in the document body, not in a Word header. .docx for career portals. Text-based PDF for direct email applications.

That's the list. Everything else is noise.

At CEEhire, every listing is checked before it goes live — whether the company can legally hire you from your country, what the contract type is, and whether the role is actually active.

The formatting advice still matters at the margins. The job being real and accessible to you matters a lot more.

Browse current openings →

Poland · Hungary · Czechia · Slovakia · Romania · Bulgaria · Croatia · Slovenia

Published: 2026-05-26 | CEEhire Blog |

CEE IT Salaries in 2026: What Developers in Hungary, Poland, Romania and the Region Actually Earn

Anikó Juhász — Wed, 17 Jun 2026 13:15:12 +0000

A senior developer working locally in CEE earns €2,800–€5,000/month in 2026. The same developer, remote for a US or UK company, earns €5,000–€10,000. Here's what engineers across the region actually take home — and what drives the gap.

The gap no one talks about enough

A senior backend developer working for a Romanian or Hungarian company earns between €2,800 and €3,500 per month in 2026. The same developer, same skills, same timezone, working remotely for a US or UK company, earns €5,000 to €10,000.

That gap is real. And it is why remote matters.

Romania

Romania's tech market is split in two. Outsourcing companies pay significantly less than product companies or direct-hire international employers.

Seniority	Local/outsourcing	Remote international
Junior (0–2 yrs)	€1,000–€1,300	€1,200–€1,600
Mid (2–5 yrs)	€1,600–€2,400	€2,400–€3,200
Senior (6+ yrs)	€2,800–€3,400	€4,000–€6,000+

Real reported Q2 2026 figures from community threads:

DevOps engineer, 2.5 years: €2,160 net
Senior JavaScript developer, 6 years, remote: €5,600 net
Backend AI/Python engineer, 4 years, B2B: €12,000 net

The last one is not a typo. AI specialization on B2B contracts produces numbers that look like Western European salaries.

Poland

Poland is the salary leader in the CEE region. Senior developers on B2B contracts invoice around €5,800–€6,300/month. AI and ML roles are growing 15–25% year over year.

Seniority	Employment	B2B
Junior	€1,700–€2,100	€2,000–€2,500
Mid	€2,800–€4,000	€3,500–€5,000
Senior	€4,000–€6,000	€5,500–€8,000+

Poland also has a significant tax advantage: the IP-box regime allows 5% income tax on qualifying software income.

Czech Republic

Seniority	Monthly (EUR)
Junior	€1,730–€2,200
Mid	€2,500–€3,500
Senior	€3,800–€4,930

Senior data scientists and ML engineers are the highest earners, with annual figures reaching €48,000+ gross.

Hungary

Seniority	Monthly (EUR)
Junior	€1,800–€2,200
Mid	€2,400–€3,200
Senior	€3,300–€4,500

Hungarian developers working remotely for Western companies can push well above the local senior ceiling.

Bulgaria

Bulgaria's local market pays below the regional average — but Bulgarian developers on B2B contracts with international clients tell a different story.

Seniority	Local (EUR/month)	Remote/B2B (EUR/hour)
Junior	€1,000–€1,400	–
Mid	€1,600–€2,300	€20–€30/h
Senior	€2,300–€3,500	€35–€46/h

At €40/h that's €6,400–€7,000/month full-time. The gap between local and international is wider here than anywhere else in the region.

What actually drives salary in 2026

Three things move salaries up across every country:

1. Who you work for. Local outsourcing → product company → direct US/UK contract. The same developer profile can see a 2–3x difference on this variable alone.

2. Contract type. B2B pays meaningfully more than employment in every country. Most senior developers in the region have already made this switch.

3. Specialization. AI, ML, cybersecurity, DevOps. Generalist roles are growing slowly or not at all.

The remote premium, by the numbers

Region	Local senior avg.	Remote international avg.	Delta
Romania	€3,000	€6,000–€8,000	+100–170%
Hungary	€3,800	€5,500–€8,000	+45–110%
Poland	€5,000	€6,500–€9,000	+30–80%
Czech Republic	€4,200	€6,000–€8,500	+43–100%
Bulgaria	€2,800	€6,000–€7,500	+115–170%

What this means if you're job hunting

The challenge isn't the salary data — it's finding the roles. Not the ones that say "remote" but require US residency. Not the ones posted six months ago. Not ghost jobs.

That's the problem CEEhire is built to solve. Every role is confirmed as accessible from the CEE region. Salary disclosed where available. Company verified as real.

Browse current listings →

Originally published on CEEhire Blog

Ghost Jobs: Why 40% of IT Job Listings Aren't Actually Open

Anikó Juhász — Fri, 12 Jun 2026 07:52:16 +0000

You spend an hour writing a cover letter. You tailor your CV. You hit submit.

Then nothing. Not a rejection. Not an automated reply. Nothing.

Two weeks later the listing is still up. You wonder if something went wrong with your application. It didn't. The job was never real.

This is the ghost job problem — and it's bigger than most people realise.

What a ghost job actually is

A ghost job is a listing that looks active but has no real intent behind it. The company isn't hiring. The role doesn't exist yet, or was already filled, or was quietly frozen after the listing went up.

Industry surveys put the scale of this at around 40%. Nearly four in ten managers have admitted their company kept an active listing live when there was no actual open position behind it.

Why companies do it

Building a talent pool. This is the most common reason, accounting for roughly 45% of ghost job postings. The company isn't hiring today but wants a pipeline for when they are. So they run a listing, collect CVs, and file them. You applied to a database, not a job.

Looking like a growing company. A careers page full of open roles signals momentum — to investors, to clients, to prospective employees. The listings themselves are part of the branding.

Keeping overworked teams quiet. A manager with a burned-out team posts a role to signal that help is coming. Then budget gets cut. The listing stays up because no one thinks to take it down.

Forgotten admin. Sometimes it's just that no one closed the listing after the role was filled. Most job boards don't expire postings automatically.

How to spot one before you apply

A real job moves. A ghost job just sits there.

The clearest signal is time. Real positions get filled in an average of 42 days. Ghost job listings stay active for an average of 180 days. If a listing has been up for two months with no changes, that's a flag.

A few others worth checking:

The listing reappears regularly. Some platforms auto-repost every 30 days. If you've seen the same role three times in four months with identical text — it's probably not moving.
The job description is generic. No tech stack. No team size. No actual role context. Written to attract CV submissions, not candidates.
The company's public signals don't match. Announced layoffs last quarter, but eight open roles on the careers page? The math doesn't add up.
No reply for weeks. Real hiring processes move imperfectly but they move. Three weeks with no acknowledgement means the listing probably isn't active.

What to do about it

Check the posting date first. Anything older than 30–45 days without an update deserves extra scrutiny.

Look at the company's LinkedIn activity. Are they actually growing? Have they posted any hiring-related content?

Find a direct contact before applying. A short message asking whether the role is still active takes two minutes and saves you an hour.

Why this matters more for CEE developers

When a company in the US or UK posts a "remote" role on a global platform, the listing gets applications from 80 countries. The company has no urgency to respond to anyone because the volume is unmanageable.

Meanwhile, a developer in Warsaw or Budapest spent an evening on that application. And the company was never going to reply regardless.

At CEEhire, one of the things I check before any job goes live is whether the listing is current and active. If the role was posted 60 days ago and shows no signs of movement, it doesn't go on the board.

That's a manual process. It's slower than running an aggregator. That's the point.

Fewer jobs. All of them actually open.

Originally published on CEEhire Blog