ATS Resume: 7 Rules That Get Yours Past the Filter (2026 Edition)

An ATS is a stricter reader than any human recruiter. A recruiter who likes you can forgive a typo, a strange company name, or a missing keyword. An ATS cannot. If your resume does not match the seven rules in this article, the recruiter never gets the chance to like you — your file is filtered out before any human opens it.

This is the pillar piece of the Bullets English series. It assumes you already know what an ATS is and how it works (if not, start with W1 — What is ATS in 2026). The seven rules below cover the practical mechanics of building an ATS-friendly resume in 2026 — under the new ATS 2.0 ranking models, with the new stuffing-detection layer, and on the specific failure modes international applicants run into.

Four of the seven rules — single-column layout, keyword matching, foreign company names, and stuffing avoidance — are where the most candidates lose points. Rules 1 and 2 are the table stakes. Rule 6 is the one nobody else explains.

---

Rule 1: Single-Column Layout, No Sidebars

The first thing a parser does is read the text from your file in top-to-bottom, left-to-right order. If your resume has a sidebar — the "modern" Canva or Word template with a colored bar on the left containing your contact info, skills, or languages — the parser has two choices: read the sidebar first and then the main column, or interleave them. Either way, the result is jumbled.

Here is what a two-column parse actually looks like to the ATS:

Priya Sharma | priya@example.com | Skills: Python, AWS, Docker
| Languages: English, Hindi | Senior Software Engineer | Infosys
| Bangalore, India | 2021-2024 | Built REST APIs | Led migration

That string is what gets matched against the job description. Your beautiful Canva template just turned itself into one long line where "Senior Software Engineer" is sitting between "Languages: Hindi" and "Infosys" — and the parser may decide your job title is "Languages: Hindi."

The rule: Use a single column from edge to edge. Header at the top with name and contact. Section headings (Summary, Experience, Education, Skills) flush left. Bullets indented under each role. No sidebar. No icons next to phone or email. No graphs of "skill levels."

If you must use a template, the ATS-safe ones are the boring ones — the Microsoft Word default resume, the LinkedIn export, or any single-column template described as "ATS-friendly" on a reputable resume site. Avoid templates marketed as "creative," "designer," or "modern." Those words are euphemisms for "two-column" in this market.

Before/after pattern:

Before (two-column)	After (single-column)
Left column: Skills, Languages, Contact. Right column: Experience, Education.	Header: Name + Contact. Then: Summary, Experience, Education, Skills, Languages — each stacked, full width.
Visual icons for phone, email, location.	Plain text labels or just the values.
"Skill bars" or rating dots for Python/SQL/etc.	A skills line: Python, SQL, Docker, AWS, Kubernetes, Terraform.

---

Rule 2: Standard Fonts, Under 1 MB File Size

Fonts and file weight cause two specific ATS failures: missing characters and rejected uploads.

Standard fonts that all ATS parse correctly: Arial, Calibri, Helvetica, Times New Roman, Georgia, Verdana. Pick one and use it for the whole document. Body text 10-11pt. Headings 12-14pt.

Fonts that cause trouble: Anything you downloaded for free. Anything that came with your design tool. Korean-only fonts like "본고딕." Hindi/Devanagari fonts that may not be embedded in the PDF. Any font with the word "Display," "Script," or "Modern" in the name.

If your name or address contains non-ASCII characters (Korean Hangul, Hindi Devanagari, Spanish accents, Vietnamese diacritics, Chinese Hanzi), make sure they are embedded in the PDF and that you also include an English transliteration. The cleanest pattern: Priya Sharma (प्रिया शर्मा) or simply Priya Sharma if your Indian passport name is what you go by professionally. For US applications, English transliteration first.

File size. Keep your resume under 1 MB. Greenhouse and Workday cap uploads at 5 MB and 10 MB respectively, but smaller is safer. Most ATS will accept a 2-3 MB file but parse it more slowly. If your resume is over 1 MB, the cause is almost always an embedded image — your profile photo, your company logos, or a background watermark. Remove them. International applicants in particular: do not include a passport-style photo on a resume going to the US, Canada, UK, Australia, or most of EU. It is illegal for the company to consider, so they will discard the resume rather than risk it.

File format. PDF for almost every modern ATS. DOCX as backup. (See Rule 5 for the full picker.) Never JPG, PNG, scanned PDF, or Pages format.

---

Rule 3: The 70% Keyword Match Rule

This is the rule most candidates think they understand and most get wrong. The pattern is simple:

Take the job description. Extract the hard skills, tools, certifications, and required titles. Your resume should contain at least 70% of those terms — naturally, in context, in your bullets — not in a separate "Keywords" block at the bottom.

How to extract JD keywords in 5 minutes:

1. Copy the full JD into a plain document.
2. Highlight every concrete noun: tools (Python, AWS, Salesforce, Figma), certifications (PMP, AWS-SAA), methodologies (Agile, Scrum, A/B testing), titles ("senior backend engineer"), and years/domain markers ("5+ years SaaS").
3. Group them by must-have (listed in "Requirements" or "Qualifications") and nice-to-have (listed in "Preferred" or "Bonus").
4. Aim for 100% of must-haves and 50%+ of nice-to-haves.

Where to put the keywords. Not in a "Skills" pile at the bottom — that is ATS 1.0 thinking. Modern ATS read context. The right place is inside the bullets describing the work you actually did. "Built CI/CD pipelines using GitHub Actions and Terraform on AWS" puts three high-value keywords in one bullet, in context.

The Skills section still has a job. Keep it, but make it the place for skills that did not naturally appear in a bullet. One line, comma-separated, no bullets, no rating bars.

ATS 2.0 nuance — synonyms. Modern parsers recognize that "REST API" ≈ "RESTful service" ≈ "HTTP API." You no longer have to write all three. Pick the term the JD uses. If the JD says "RESTful APIs," write "RESTful APIs."

A worked example. Suppose the JD says: "5+ years Python, experience with FastAPI or Flask, AWS (Lambda, S3, RDS), Docker, Kubernetes, CI/CD with GitHub Actions, monitoring with Datadog or New Relic, agile/scrum."

That is 11 hard terms. A 70% match means at least 8 of them must appear in your resume — and the ATS 2.0 parser is also checking that those terms appear next to verbs that describe doing that work, not under "Skills: Python, FastAPI, …" alone.

The Bullets Keyword Mapping tool is built exactly for this step — paste your resume + JD URL, get the missing-keyword list, and get bullet-level rewrite suggestions where each missing term naturally fits.

---

Rule 4: Start Every Bullet with an Action Verb

The phrase "Responsible for" is the most common opener on weak resumes and the strongest signal to an ATS that this bullet is vague. ATS 2.0 ranking models down-weight bullets that start with "Responsible for," "Duties included," "Helped with," or "Assisted in." They up-weight bullets that start with concrete verbs tied to outcomes.

The opening verb is also the keyword. "Led" is a leadership signal. "Reduced" is a metrics signal. "Built" is an IC signal. "Migrated" is a project-completion signal. The recruiter's eye and the model both lock onto that first word.

Categorize your bullets by which verb to use:

• Built / Designed / Shipped / Implemented / Developed — IC work that produced something.
• Led / Managed / Mentored / Coached — people leadership.
• Reduced / Increased / Improved / Optimized / Accelerated — outcome bullets with numbers.
• Migrated / Refactored / Consolidated / Standardized — large-scope cleanup work.
• Designed / Drove / Owned / Spearheaded — strategic ownership.
• Won / Closed / Negotiated / Landed — sales and BD.

Pick the verb that best matches the actual work. Do not write "Led" if you did not lead — ATS 2.0 cross-checks with context, and a "Led" bullet that doesn't mention team size or scope looks suspicious.

Bad → good rewrites:

Bad (vague)	Good (action + outcome)
Responsible for backend services.	Built and deployed 12 microservices in Python/FastAPI, serving 4M requests/day.
Worked on the data team.	Designed the company's first event-tracking pipeline on Snowflake; reduced reporting latency from 24h to 15min.
Helped with code reviews.	Reviewed 200+ PRs/quarter as senior reviewer; cut average PR-to-merge time by 35%.

Read W5 — STAR Method + PMTVQ (publishing later in this series) for the full verb + quantification framework for converting non-US experience into US-shaped bullets.

---

Rule 5: PDF vs DOCX — Which Is Safer?

The short answer: PDF in 2026, with two exceptions.

The long answer:

• Modern ATS (Greenhouse, Lever, Workday post-2022, SmartRecruiters, BambooHR, Ashby) parse PDF and DOCX equally well. PDF preserves your formatting so the human-readable version matches what you intended. PDF is the default choice.
• Legacy ATS (older Taleo, iCIMS installations at large enterprises and government) sometimes parse DOCX better than PDF. If you are applying to a Fortune 500 with a careers portal that looks like it was built in 2012, DOCX is safer.
• If the job posting specifies a format, follow it. "Submit your resume as a Word document" means DOCX. Submitting PDF anyway is a small but real signal that you do not read instructions.

A practical decision tree:

Situation	Format
The JD specifies PDF or DOCX.	Match what they ask.
The application portal looks modern (Greenhouse/Lever/Ashby URL, clean UI).	PDF.
The portal is a generic-branded careers page with a 2012 feel.	DOCX.
Applying to a US government, university, or large-bank role.	DOCX.
Applying to any tech startup, scaleup, or FAANG-tier company.	PDF.
Submitting via LinkedIn Easy Apply.	PDF.
Sending directly to a recruiter by email.	PDF.

How to save a PDF that parses cleanly. Export from Word or Google Docs as PDF. Open the PDF and try to highlight and copy a sentence from it. If the highlighted text copies as actual text, you are fine. If it copies as garbage or only one character, your PDF is image-based and will fail every parser. Re-export.

Never use:
- Scanned PDFs.
- PDFs exported with "fonts as outlines" or "all as image."
- The .pages format from Apple Pages — half of ATS reject it outright.
- LinkedIn's "Save to PDF" of your profile as your only resume — it parses, but the formatting is generic and the keyword density is low.

---

Rule 6 (International-Specific): Foreign Company and School Names

This is the rule that nobody else writes about, and it is where Indian, Korean, LATAM, and Southeast-Asian candidates lose the most ground in 2026.

The ATS 2.0 ranking model uses company name as a feature. A "Software Engineer at Google" outranks a "Software Engineer at Acme" for the same JD because the model has been trained on signals that say Google → high-bar engineering culture. That model has also been trained almost entirely on US data. If your last company is TCS, Infosys, Wipro, Globant, Mercado Libre, Coupang, Grab, GoTo, or Samsung SDS, the model may not know what those names mean. Worse, the parser may not even recognize them as company names — it may classify them as projects or skills.

The fix pattern is one extra line per non-US company:

Tata Consultancy Services (TCS) — India's largest IT services firm, 600,000+ employees
Senior Software Engineer | Bangalore, India | Aug 2021 – Mar 2024

That descriptor line does three jobs:

1. Tells the parser this is a company name (because "company" patterns like "the largest X firm" trigger the right classifier).
2. Tells the recruiter the scale and credibility of your previous employer.
3. Adds keywords (IT services, large-enterprise, scale) without stuffing.

Examples by region:

• Infosys — Indian IT services major, NYSE-listed, 300,000+ employees
• Samsung SDS — Samsung Group IT subsidiary, global enterprise IT integrator
• Coupang — South Korea's largest e-commerce platform, Nasdaq-listed
• Rappi — Latin America's largest super-app, valued $5B
• Globo — Brazil's largest media conglomerate, owner of TV Globo and Globoplay
• Grab — Southeast Asia's super-app leader, Nasdaq-listed

Schools follow the same pattern, but only for non-famous schools. If your school is IIT, IIM, KAIST, SNU, NUS, Tsinghua, or USP — most US recruiters and most ATS 2.0 models already know them. A short tag still helps but is not critical. For NIT Trichy, BITS Pilani, Anna University, Korea University, ITESM, FGV, or any regional university, add the descriptor:

National Institute of Technology (NIT) Trichy — top-tier Indian engineering school, B.Tech Computer Science, 2021

The same logic applies to in-country promotions. If you were promoted twice at the same Indian or Korean company over four years, write each title and the descriptor of the role, not just the dates. The parser otherwise reads four years of "the same job."

Infosys — Indian IT services major, NYSE-listed, 300,000+ employees
  Senior Software Engineer (SE-III) | Apr 2023 – Mar 2024
  Software Engineer (SE-II)         | Jul 2021 – Mar 2023

This pattern is also covered in depth for the H-1B persona in W3 — H-1B Visa Resume: From Indian Engineer to US Big Tech.

---

Rule 7: Avoid ATS Keyword Inflation

Keyword stuffing used to work. In ATS 1.0, repeating "Python" 12 times across your resume would raise your match score. In ATS 2.0, the model actively detects repetition and stuffing patterns and penalizes the resume. The current rule of thumb:

Each important keyword should appear 2-4 times across your resume. Once or twice is normal usage. Three or four times across different bullets is reinforcement. Five or more is a stuffing signal.

What stuffing looks like (and gets caught):

• A 10-line "Skills" section listing 80 technologies, half of which never appear elsewhere.
• The same word ("Python") repeated as the first word of five consecutive bullets.
• White-on-white text with keywords behind the visible content. (All major ATS strip this and flag it.)
• A keyword block at the bottom titled "Additional Skills" with another 50 terms.

What healthy keyword density looks like:

• "Python" appears in your summary line, in two specific bullets describing what you built, and once in your Skills line — 4 occurrences, all in context.
• Synonyms used naturally: "REST APIs" in one bullet, "RESTful services" in another.
• The Skills section is 8-12 items, not 50.

ATS 2.0 stuffing detection has gotten precise. Greenhouse, Workday, and Lever shipped LLM-assisted stuffing detection between 2024 and 2025. The model looks for:

• High keyword density without contextual bullets.
• Skills listed that never co-occur with verbs.
• Repetition patterns inconsistent with natural writing.

If your resume gets flagged, the consequence is not just a low score — some systems route flagged resumes straight to a "review" queue that recruiters rarely open.

Synonym strategy. Instead of repeating one term, alternate among 2-3 industry-standard synonyms. For the term "machine learning," you can mix "ML," "machine learning," "predictive modeling," and "model training" across different bullets. The parser recognizes them as related. The model rewards the natural variation.

---

Closing: How to Check Your ATS Score in One Minute

You have read seven rules. Reading them is not the same as having a resume that passes. The fastest way to find out where your file currently fails is to run it through an ATS-style scanner with a target job description.

The Bullet Creator tool from Bullets does this in under a minute:

1. Upload your current resume (PDF or DOCX).
2. Paste a job description URL or text.
3. Get a score for each of the 7 rules:
4. Parsing health (single-column, fonts, file size).
5. Keyword match % vs the JD.
6. Action verb coverage and weak-verb flags.
7. Quantification rate (% of bullets with numbers).
8. International-name descriptor coverage.
9. Stuffing risk score.

The tool also returns bullet-level rewrite suggestions in the STAR + PMTVQ format used across this series.

If you are an Indian engineer specifically targeting H-1B-sponsoring companies, the next article is built for you: W3 — H-1B Visa Resume: From Indian Engineer to US Big Tech. It covers the H-1B process, the 5 persona-specific failures, visa signaling language, and a free Top 100 H-1B Sponsoring Companies 2026 PDF.

For the basics — what an ATS actually is and how it works — go back to W1 — What is ATS in 2026.

---

Sources: Jobscan ATS Resume Report (2024), Greenhouse and Lever public product documentation, Workday Talent Cloud release notes (2024-2025), LinkedIn Talent Insights, Bullets internal analysis of 1,800+ resume-to-JD scoring runs (2024-2026).