Why You're Not Getting Interviews: The ATS Filter You Didn't Know Existed

You have applied to forty roles in the last month. Some you were genuinely qualified for. A handful you were probably overqualified for. And the response rate has been, what — two automated rejections and twenty-eight cases of total silence?

You are not doing anything obviously wrong. Your CV looks fine. Your cover letter is competent. Your experience matches the job description. So what is happening?

Almost certainly, this: the applicant tracking system rejected your CV before a human ever opened it. And it did so on the basis of criteria that have very little to do with whether you are good at the job, and very much to do with whether your CV happened to be machine-readable in the way the ATS expected.

This is not a conspiracy theory and it is not new. It is, however, much less understood than it should be. This piece walks through what actually happens to your CV after you click submit, why so many qualified candidates get silently filtered out, and a concrete checklist for getting your CV through the gate.

The Three Things That Happen When You Click Submit

Most candidates think of "applying" as a single action. From the ATS's perspective, three things happen in sequence:

1. Parsing — your CV, in whatever format you uploaded, gets fed through a parser that tries to extract structured data: name, contact info, work history, education, skills.
2. Scoring — the parsed data is compared against the job description, usually via keyword matching, sometimes via more sophisticated semantic comparison.
3. Filtering — the recruiter sees a ranked list, often with a cut-off, and reviews only the top N candidates. For popular roles, N is usually 20-50 out of several hundred applicants.

The interesting question is what fraction of qualified candidates fall out at each stage. From conversations with recruiters and ATS engineers, the rough answer is:

• ~15-30% of CVs fail parsing badly enough that the recruiter never sees them, or sees them with critical fields missing.
• ~40-60% of CVs that parse successfully fail to clear the keyword threshold for the role.
• The remaining 20-30% are the pile the recruiter actually reads.

If you have been wondering why you do not hear back, the answer is statistical. Even a strong CV has, at best, maybe a 25% chance of making it to a human eye on any given application. And that is before the human filters further.

How ATSes Actually Match Keywords

The "keyword match" idea has been around for years, but it is poorly understood. Here is the actual mechanism on the most common platforms.

When a job is posted, the recruiter or hiring manager fills out a form that includes the job description. Behind the scenes, the ATS extracts a list of "skills" and "requirements" from that text — sometimes manually entered by the recruiter, sometimes pulled by NLP. This becomes the scoring rubric for incoming applications.

When your CV arrives:

1. The ATS extracts a similar list of skills and experience markers from your CV.
2. It compares your list to the rubric.
3. It produces a score — typically 0-100, sometimes presented as a star rating or "match percentage".
4. The recruiter sorts by that score.

Critical detail: the comparison is more literal than you would expect. "Project management" and "PM" are often treated as different keywords. "Python" and "Python 3.9" sometimes are too. "Managed a team of five" can match "team management" or it might not, depending on the parser.

The implication is uncomfortable: tweaking the wording of your CV to match the language of the job description, almost line by line, can move you up the score pile substantially. This is not gaming the system — it is reading the system correctly. The job description is the rubric you are being graded against.

Why Parsing Fails (And What That Looks Like To You)

A CV "failing to parse" does not mean the ATS rejects it outright. It means the ATS fills the structured fields with whatever it could grab, even if that data is wrong, and then scores the broken result.

The classic failure modes:

Two-column layouts. Most modern ATS parsers are better than they used to be at handling two columns, but the results are still inconsistent. Some parse left-to-right, top-to-bottom across both columns, producing a jumbled mess. Some parse one column then the other, getting it right by accident. Some give up.

Headers and footers. Your contact info is in the header? Some parsers ignore headers entirely. Your name is in the footer? Same problem.

Custom fonts and icons. Font Awesome icons next to your phone number may render as a question mark or get extracted as a literal Unicode character that the ATS does not understand. Custom fonts sometimes break ligatures so "fi" becomes "?".

Tables. Tables embedded in Word documents often parse as one long horizontal row of data. Your beautifully tabulated work history becomes "Senior Engineer Acme Corp 2019-2022 Built things Frontend Engineer Acme Corp 2017-2019 Built other things" and the ATS has no idea where one job ends and the next begins.

Graphics and images. Skill bars, charts, infographics — invisible to the parser. If your "skills" section is a chart of bar lengths, the ATS extracts no skills at all.

PDFs from design tools. A PDF exported from Canva or Figma is sometimes a flattened image with no extractable text. The ATS sees a blank page. We have personally seen this in the wild and it is brutal.

Mismatched section headings. Most parsers look for standard headings: "Experience", "Work Experience", "Employment History", "Education", "Skills". If you used "My Journey" or "What I've Done", the parser may skip the section entirely.

Dates in odd formats. "Spring 2022 - Present" is harder to parse than "March 2022 - Present". Date ranges with em-dashes (–) versus hyphens (-) versus tildes (~) are inconsistently handled.

The unsettling thing is that most of these failures are silent. You will not get an error message. The ATS will quietly fail to extract your work history and score you as if you had no relevant experience.

The Difference Between Lever, Greenhouse, Workday, and iCIMS

Not all ATSes are the same. Knowing which one you are facing changes how you approach the application.

Lever

Used by: many tech startups, some scale-ups (Stripe historically, Brex, plenty of YC companies).

Strengths: Clean candidate experience, fast applications, recruiter UI is genuinely good. Parsers are reasonably forgiving — Lever handles most modern CV formats acceptably.

Filtering style: Less aggressive automated filtering. Recruiters usually look at most applications, especially for early-stage companies. The CV-vs-keyword score matters less. Cover letter and notes matter more.

Optimisation tip: Plain CV, then put effort into the application questions and cover letter. Lever-using companies tend to read.

Greenhouse

Used by: many mid-size to large tech companies (Airbnb, Pinterest, Stripe currently, hundreds more).

Strengths: Sophisticated workflow management, structured interview kits. Greenhouse has matured into the dominant tech-company ATS.

Filtering style: Mixed. Greenhouse provides keyword scoring tools but how much weight a recruiter gives them varies enormously. Some Greenhouse companies use it heavily, some treat the score as an FYI.

Optimisation tip: Match key skills from the JD verbatim. Greenhouse-using companies tend to have well-written job descriptions, so the rubric is clearer than at companies using older systems.

Workday

Used by: enterprise. Banks, retailers, universities, government contractors, healthcare systems. Most Fortune 500 companies have at least some recruiting on Workday.

Strengths: Integrated with the rest of Workday's enterprise HR suite, so the data flows nicely once you are hired.

Filtering style: Heavier automated filtering. Large companies receive thousands of applications per role and need to cut the pile aggressively. Keyword match score matters a lot. So does completeness — if you skipped the optional fields, you may rank lower than a complete (if otherwise weaker) candidate.

Optimisation tip: Maximally ATS-friendly CV, complete every field, paste your job description keywords directly into your skills section. We covered the specific mechanics in our Workday application guide.

iCIMS

Used by: many large non-tech companies. Healthcare, retail, manufacturing, logistics.

Strengths: Highly customisable for the recruiting team, handles complex regulatory requirements (drug testing, background check workflows).

Filtering style: Variable. iCIMS has the most customisable scoring in the industry, which means each company configures it differently. Some screen heavily on years-of-experience, some on certifications, some on degree level. There is no general optimisation rule beyond "match the JD".

Optimisation tip: Read the JD twice. iCIMS often hard-filters on stated requirements ("must have CPA", "5+ years required"), so if the JD says it, your CV needs to say it explicitly.

Others worth mentioning

Taleo (Oracle) — older system, used by many large legacy enterprises. Parser is genuinely bad. Plain CV, single column, repeat keywords explicitly.

SuccessFactors (SAP) — similar to Workday in audience, slightly better parser. Same advice applies.

ApplyBe — increasingly common in mid-market; modern parser, candidate experience is reasonable.

SmartRecruiters — mid-market, good candidate experience, parser is OK but less forgiving than Lever's.

If you are not sure which ATS a company uses, look at the URL of the application page. Lever applications live on jobs.lever.co/[company]. Greenhouse on boards.greenhouse.io/[company]. Workday on [company].wd1.myworkdayjobs.com. iCIMS on careers-[company].icims.com. Taleo on [company].taleo.net.

A Checklist for ATS-Proofing Your CV

Here is the concrete list. None of this is glamorous. All of it works.

Format

• Single column layout, full width
• Standard fonts only: Arial, Calibri, Helvetica, Georgia, Times New Roman
• 11pt or 12pt body text (10pt minimum)
• No tables for layout (some tables are OK; avoid for work history)
• No headers or footers — put contact info in the body
• No graphics, charts, skill bars, icons, or coloured backgrounds
• No text in images (the ATS cannot read it)
• Standard section headings: "Experience", "Education", "Skills"

Content

• Contact section at the top: Name, Email, Phone, City+Country, LinkedIn URL
• Each role on a new line: Job Title | Company | Location | Date Range
• Date format: "January 2022 - Present" or "01/2022 - Present" (consistent)
• Skills section with 8-15 explicit skills, named the way the job description names them
• Education section with degree, institution, and year
• No personal pronouns ("Built X" not "I built X")

Keyword targeting

• Read the job description carefully. Underline 10-15 specific terms.
• Make sure each appears in your CV at least once, ideally in the section a recruiter would expect.
• If you have a relevant certification, list it explicitly even if "obvious" (e.g. "AWS Certified Solutions Architect" not "AWS certified")
• If the JD lists tools (e.g. "Jira, Confluence, Notion"), name them in your skills section
• Avoid abbreviations the JD does not use; spell things out

File format

• Save as .docx for ATS uploads (PDFs work but parse less reliably)
• Filename should be Firstname_Lastname_CV.docx
• No fancy file metadata, no password protection
• Test by copying all text out of the file with Ctrl+A, Ctrl+C and pasting into Notepad — that is approximately what the parser sees

Final test

• Run your CV through a free ATS scanner (Jobscan, Resume Worded, several others)
• If the scanner cannot extract your work history or contact info, neither can the ATS
• Make a separate "ATS-safe" version of your CV. Keep your designed version for direct contacts and LinkedIn.

If this list seems boring and unimaginative, that is the point. The ATS rewards CVs that look like every other ATS-friendly CV. There is no aesthetic prize for getting through the filter.

The Myth of Creative Resume Design

A persistent piece of advice on LinkedIn and TikTok is to make your CV "stand out" with creative design — colour, columns, graphics, custom fonts. The argument is that recruiters are buried in lookalike CVs and yours needs to catch the eye.

This advice is well-meaning and badly wrong, at least for any role at a company that uses an ATS. Two reasons:

1. The recruiter does not see your beautiful design. They see the parsed output in their ATS dashboard. Your two-column layout becomes a single column of text in a generic template. Your custom fonts become Arial. Your skill bars become missing data. The design effort is invisible to the audience it was meant for.

2. Even if the recruiter opens your original CV, they spend on average less than 10 seconds on it. Recruiters tell us they look for: the most recent role and company, the dates, the skills section, and one or two notable items in the experience. A creative layout makes those harder to find, not easier.

The exceptions are real but narrow:

• Senior creative roles (Design Director, Creative Lead) where the CV doubles as a portfolio piece and the recipient will look at it directly, often outside an ATS.
• Direct introductions where someone forwards your CV to a hiring manager. No ATS in the loop.
• Founder-led startups where the CV may go straight to the CEO's inbox.

For everything else — and that is the vast majority of applications — boring CV beats beautiful CV. By a lot.

How Multi-CV Testing Reveals What Actually Works

Here is the most useful diagnostic exercise we know of: maintain three versions of your CV, submit all three to similar roles over a few weeks, and track which one gets the best response rate.

A typical setup:

• Version A: Your "designed" CV — the one you have been using.
• Version B: A plain, single-column, ATS-optimised version of the same content.
• Version C: Same as B but with the skills section rewritten to match the language of the specific job descriptions you are applying to.

Apply to comparable roles with each version and track the response rate over 2-4 weeks. The pattern is almost always the same: B beats A by 30-50%, and C beats B by another 20-30%. The total uplift from "designed CV with my normal phrasing" to "plain CV with JD-matched phrasing" is often 2-3x in interview rate.

This is the practical case for AI-driven CV scoring. If you can score your CV against a job description before you apply — see which keywords are missing, which sections look weak to a parser — you can fix the issues that would have rejected you, in advance. This is exactly what tools like ResuMinder do: they score every CV you have uploaded against every job, surface the gaps, and let you tailor before you click submit instead of after the silent rejection.

The principle generalises beyond any specific tool. If you are applying to roles without first checking how your CV scores, you are flying blind. Even doing the comparison manually — spending five minutes lining up the JD's keywords against your CV's text — beats no comparison at all.

What To Do This Week

If you are reading this and recognising yourself, here is an honest action list for the next seven days.

Day 1: Save a fresh copy of your current CV. Open it in Word and reformat to the ATS-safe spec above. Single column, plain fonts, standard headings, contact info in the body. Save as Firstname_Lastname_CV.docx.

Day 2-3: Pick three job descriptions for roles you would actually want. For each, list the 10-15 most-mentioned skills and tools. Compare against your CV. Where the CV is missing terms, add them — but only where genuinely true. "Familiar with X" is fine if you are; "expert in X" is not unless you are.

Day 4: Run all three tailored versions through a free ATS scanner. Note the parse score and any critical extraction failures (missing contact info, missing work history). Fix what is broken.

Day 5: Apply to 5-10 roles using the new CVs. Track which version goes to which role.

Day 6-7: Rest. Then review responses over the following two weeks and compare to your previous baseline.

If the response rate does not improve, the bottleneck is not your CV — it is probably keyword match (apply to roles closer to your actual experience) or volume (you may need to apply more, full stop). But for a substantial fraction of job seekers, the simple act of switching to an ATS-friendly format produces a measurable bump within two weeks.

The Honest Conclusion

The ATS filter is real. It is not malicious, it is not particularly smart, and it is not going away. It exists because companies receive too many applications and need to triage. The candidates who get interviews are not necessarily the best candidates — they are the candidates whose CVs the ATS thinks are the best.

You can be furious about this or you can be practical about it. The practical version: format your CV for the parser, match the JD's language, test before you apply, and apply more rather than less. None of this is creativity. All of it works.

If you would like the testing and scoring done automatically — comparing every CV you have to every job description, surfacing the missing keywords, and ranking your CV options against each role — that is what ResuMinder was built for. You can compare it against alternatives on our comparison page, or read about the features before you decide.

Whichever tool you use, the underlying lesson is the same. Stop applying with the same CV to every role and assuming the recruiter will see your potential. The recruiter has not seen your CV yet — the ATS has. Convince it first.

---

For deeper reading on the keyword and formatting side of this problem, our ATS resume optimisation guide breaks down the specific scoring criteria most platforms use. And if you are wondering whether automation is worth it for the volume side, our automation guide goes through the maths.