If you have ever opened a blank document to write your resume, you have probably wondered why resume advice sounds so strict. Why can’t you put your email in a stylish header box at the top of the page? Why can’t you call your work history “Where I’ve Been” or show your Python skills with a five-star graphic bar?
If only a human read your resume from a printed sheet of paper, many of those creative choices would merely be a matter of taste. In modern hiring, however, your file is read by two completely different readers before you ever get an interview call:
- Reader 1 — The Software Extractor (Applicant Tracking System): The moment you upload a PDF or Word document to a company portal (such as Workday, Greenhouse, iCIMS, Lever, SAP SuccessFactors, or Oracle Taleo), an extraction program reads the underlying text stream of your file. Its job is to slice your document into database fields:
Candidate Name,Email,Phone,Location,Work History,Role Titles,Start & End Dates,Education, andSkills. - Reader 2 — The Human Reviewer (Recruiter & Hiring Manager): Once your file is indexed, a recruiter views either the structured profile or a side-by-side preview of your page. On a first pass, they do not read your resume like a novel from top to bottom. They scan the top third of page one — your current title, your recent companies, your dates, and the first line of each bullet — quickly, to decide whether to read closer.
The Guul ATS Golden Standard is our research-backed specification for writing and structuring a resume so both readers get exactly what they need on the first pass — without superstition, guesswork, or ugly plain-text compromises.

Why a resume cannot be built like a poster
When a designer builds a poster, a slide deck, or a restaurant menu, only the visual surface matters. If two text boxes sit next to each other and look aligned to the human eye, the poster works.
A resume file is different. Inside a PDF or .docx file, there is a visual layer (where shapes, columns, and fonts appear on screen) and a logical text stream (the sequence in which characters and lines are stored inside the file).
When a resume breaks the expectations of that text stream, three things go wrong quietly:
- Silent field loss during autofill: If your contact details sit inside a repeating Word document header or a floating graphic shape, platforms whose parsers skip header/footer streams leave your email or phone blank in the candidate database.
- Scrambled lines across columns: If a two-column layout places a narrow skill list at the exact same vertical height as a work-experience line without proper block isolation, coordinate-based PDF extractors read straight across the page and splice the two columns into a single garbled sentence.
- Zero-month experience calculations: If you write dates casually (such as "Summer '22 – Winter '24"), date normalizers cannot compute your tenure. When a recruiter filters the candidate pool for "3+ years of experience", a three-year role with unparseable dates can register as zero months.
Use the interactive blueprint below to switch between a fragile resume and the Guul Golden Standard, read what five major hiring platforms say in their own documentation, or run the 8-point self-audit on your own resume.
See your resume through both eyes: the software parser and the human scan.
Explore all six rules of the Guul ATS Golden Standard below. Switch between a fragile resume and the Golden Standard to watch the live parser scan beam and the recruiter’s eye-path, read what five major hiring platforms say in their own documentation, or run the 8-point self-audit on your own file.
Place your name, email, phone, and city in the main document body.
Why this rule exists
When you double-click the very top edge of a Word page, you enter the repeating page header. It looks normal on screen, but many hiring portals ignore repeating headers and footers so page numbers do not pollute your work history. Putting your contact line at the top of the normal page body keeps you reachable.
What parsers and hiring systems actually do
In OOXML (.docx), headers live in separate word/header1.xml parts outside word/document.xml. Greenhouse’s own support documentation lists contact details placed in a header, footer or text box among the causes of an unsuccessful resume parse. In PDFs, icon fonts (Private Use Area glyphs) and unspaced punctuation can also break regex phone/email tokenizers.
What actually happens after you click Submit
Before diving into the six rules, it helps to replace internet folklore with how hiring software actually works.
Note: Hiring platforms do not run a secret "visual beauty judge" that rejects resumes for using a shade of green or a two-page document. Published documentation from Workday, Greenhouse, Lever, SAP SuccessFactors and Oracle Taleo describes a much more mechanical pipeline.
Step 1: File intake and text-layer extraction
When you upload a .pdf or .docx file, the system first extracts raw text:
- For
.docxfiles, the parser reads the OpenXML document structure (word/document.xml). Paragraphs, headings, and lists are stored in sequential nodes, while repeating page headers (word/header1.xml), footers, and floating VML/DrawingML shapes sit outside the main body flow. - For
.pdffiles, the parser reads positioned glyph runs(x, y, text)from the PDF content stream. Some extractors group glyphs into layout blocks first (likePyMuPDF), while simpler physical-layout extractors (likePoppler pdftotext -layout) sort characters top-to-bottom and left-to-right across each horizontal line. If your PDF is a flat image scan with no selectable text layer, extraction returns nothing unless the employer runs optical character recognition (OCR).
Step 2: Section segmentation and field mapping
Next, the parser searches for section boundaries. It looks for short, isolated heading lines that match its known vocabulary: Summary, Experience / Work History, Education, Skills, Projects, and Certifications.
Once it identifies the Work Experience block, it looks for repeating entry anchors: a Job Title, an Employer Name, a Location, and a Start – End Date Range. Everything between one job header and the next is attached as the description of that role.
Step 3: Knockouts, indexing, and recruiter search
How do candidates actually get filtered before a human reads their resume?
- Application knockout questions: Automated rejections come from rules an employer sets on application questions, for example work authorization, a required license or location. Oracle Taleo's documentation calls these disqualification questions: single-answer questions that hold a job's minimum requirements, whose answers decide whether a candidate moves forward or is automatically disqualified.
- Structured field & keyword filters: When hundreds of people apply, recruiters search the database by structured fields (Location, Years of Experience, Current Title) and boolean keyword queries (
Python AND (PostgreSQL OR BigQuery)). If your skills were drawn as graphic progress bars or your dates failed to parse, your profile can miss those filters even though the visual PDF looks fine.
The 6 rules of the Guul ATS Golden Standard
Every formatting recommendation we make reduces to six concrete rules. Each rule solves a specific failure mode in software extraction while making the page faster for a human reviewer to scan.
Rule 01: Put your contact block in the main page body
Your full name, email address, phone number, city/state (or country), and portfolio/LinkedIn URL belong at the very top of page one inside the normal document body.
- For first-time job seekers: Do not double-click the top margin in Microsoft Word or Google Docs to type your name and email into the repeating page header. Write them as the first lines of the normal page, separated by plain middle dots (
·) or vertical bars (|). Write out the text of your email and LinkedIn URL instead of hiding them behind small envelope or icon graphics. - For advanced readers: Greenhouse's support documentation lists contact details placed in a header, footer or text box among the causes of an unsuccessful resume parse. In our extraction study, Mozilla
PDF.jsreturned substituted characters for the phone number and the date range in all 20 PDFs, although both were typed in plain text, while PyMuPDF and Poppler read the same files correctly. Check the text your exported file actually gives.
Rule 02: Use literal, standard section headings
Label your sections with the exact words both hiring software and human recruiters expect:
| Section Purpose | Golden Standard Heading | Avoid (Breaks Section Mapping) |
|---|---|---|
| Career overview | Summary or Professional Summary |
Me in 30 Seconds, Philosophy, About |
| Jobs & internships | Work Experience or Experience |
Where I've Been, Career Journey, Impact |
| Degrees & schools | Education |
Academic Background & Credentials |
| Technical & core skills | Skills or Technical Skills |
My Superpowers, Toolbox, Stack |
| Portfolio & capstones | Projects |
Things I Built, Lab Work |
| Licenses & credentials | Certifications |
Badges & Accolades |
- For first-time job seekers: Creativity belongs in the accomplishments you describe, not in the labels of your sections. Standard headings act like highway signs so a busy reviewer never has to guess where your jobs end and your education begins.
- For advanced readers: Section classifiers rely on dictionary matching plus line-level typographic cues. When an unrecognized heading appears, the parser either merges the following entries into the preceding section (attaching your university degree to the bottom of your earliest job) or dumps the text into an unindexed raw notes field.
Rule 03: Maintain a linear top-to-bottom reading stream
Whether you choose a classic single-column layout or a structured sidebar layout, the extracted text stream must read one complete thought at a time.

- For first-time job seekers: A single-column layout (
Crisp,Meridian) is the safest default, because full-width lines leave an extractor only one order to read them in. If you prefer a sidebar layout (Onyx) to fit contact details and skills cleanly on one page, always test the exported PDF by selecting all text (Cmd/Ctrl + A), copying, and pasting into a plain text editor to confirm lines from the left column do not paste into the middle of sentences on the right. - For advanced readers: Different PDF engines reconstruct reading order differently. In our extraction study,
PyMuPDFand MozillaPDF.jskept the declared section order in all 20 PDFs, while Poppler'spdftotext(default reading-order settings) mixed the two columns in all 10 sidebar PDFs. Always check your exported PDF with our Resume PDF Text Check.
Rule 04: Pair every role and degree with explicit Month + Year dates
Every work experience and education entry needs four distinct pieces of data in a predictable pattern: Role Title, Company / Institution, Location, and Start Date – End Date.
Golden Standard Role Header: Senior Data Analyst · Shopify · Ottawa, ON
May 2022 – Present
- For first-time job seekers: Always write the month and all four digits of the year (
May 2022 – Nov 2024or05/2022 – 11/2024), and usePresentorCurrentfor a role you still hold. Never write seasons (Summer 2023), quarters (Q3 2022), or two-digit apostrophe years ('21 – '24). - For advanced readers: Experience-duration calculators parse start and end tokens into
YYYY-MMintervals and sum non-overlapping months per skill or role family. Missing months or seasonal words can make a parser guess the month or drop the dates entirely, and a role without parsed dates adds nothing to a years-of-experience filter.
Rule 05: Write skills as plain-text words in context
List your technical skills, tools, methods, and spoken languages as plain text grouped by clear categories — and repeat your core skills inside the experience bullets where you used them.
- For first-time job seekers: Delete every progress bar, pie chart, star rating, or percentage meter (
Python — 90%). No recruiter knows what "80% Excel" means, and software deletes the graphic bars completely. Instead, group your skills clearly:Languages: Python, SQL, TypeScript · Analytics: Looker, dbt, BigQuery. - For advanced readers: Both structured skill extraction and recruiter keyword search operate on text tokens. Repeating a skill inside a dated work-experience bullet connects that skill to a specific employer and time window, which matters when recruiters review search highlights to verify recent hands-on production use rather than a bare keyword list.
Rule 06: Write every bullet as Action + System/Scope + Measured Outcome
Once Rules 01 through 05 make clean parsing far more likely, Rule 06 wins the human decision.
Most weak resumes read like a job posting: they list what the person was supposed to do ("Responsible for managing customer onboarding and weekly reporting"). Two candidates with the exact same job title can have wildly different performance; duty bullets hide the difference.
Every bullet in the Golden Standard answers three questions in one sentence:
- Action Verb: What specific action did you take? (Redesigned, automated, negotiated, migrated, reduced, launched)
- System or Scope: What product, workflow, dataset, or customer segment did you touch, and with which tools?
- Measured Outcome: What changed in the real world as a result? (Time saved, latency reduced, revenue grown, error rate dropped, volume handled)
Same story. A little more breathing room.
Switch the layout. Watch how spacing and hierarchy change the way this sample reads.
Alex Morgan
Product designerProfile
Product designer focused on making complex workflows easier to understand.
Experience
Product Designer · Example Studio
- Mapped the onboarding journey with the support team.
- Designed and tested a clearer account setup flow.
- Documented reusable components for the engineering team.
Skills
Interaction design · Prototyping · Design systems
Everything is asking for attention.
Closely packed headings and text leave fewer visual signposts. Try the other treatment.
Before and after: three example rewrites
Here is how the Guul ATS Golden Standard transforms typical resume entries across three different career stages. The people and numbers are illustrative.
1. First-time job seeker / recent graduate
Fragile version (hard to parse, duty-focused):
Intern (Summer '24) @ FinTech Co
- Helped the marketing team with social media campaigns and email newsletters.
- Used Excel and SQL to look at customer sign-up numbers.
Guul Golden Standard version: Growth Marketing Intern · FinTech Co · New York, NY
Jun 2024 – Aug 2024
- Built a weekly cohort tracking workbook in SQL and Microsoft Excel across 18,400 trial sign-ups, identifying a verification step where 31% of mobile users dropped off.
- Drafted and A/B tested 6 onboarding lifecycle emails in HubSpot, lifting 7-day trial-to-active conversion from 14.2% to 19.8%.
2. Experienced software & infrastructure engineer
Fragile version (vague scope, missing dates & metrics):
Backend Developer — CloudScale ('21–'24)
- Responsible for payment microservices, Kafka consumers, and database optimization.
- Worked on on-call incidents and improved system reliability.
Guul Golden Standard version: Senior Backend Engineer · CloudScale · Austin, TX
Mar 2021 – Oct 2024
- Re-architected idempotent payment settlement consumers in Go, Apache Kafka, and PostgreSQL, reducing p95 checkout latency from 480 ms to 185 ms across 1.4M daily transactions.
- Introduced automated dead-letter replay and schema validation in Kubernetes, cutting weekly on-call payment pages from 11 incidents/week to under 1.
3. Operations, finance, or product lead
Fragile version (buzzword-heavy, no concrete proof):
Operations Manager | Global Logistics
- Spearheaded cross-functional operational excellence initiatives and vendor management across regional warehouses.
Guul Golden Standard version: Regional Operations Manager · Global Logistics · Chicago, IL
Jan 2020 – Present
- Consolidated 14 regional freight carriers into 5 SLA-backed contracts across 3 Midwest fulfillment centers, lowering outbound shipping cost per order by 12.4% ($640K annual savings) while improving on-time delivery from 91% to 97.6%.
What five hiring platforms say in their own documentation
We do not guess how applicant tracking systems work. Below is what five widely used platforms document, as recorded in Guul's hiring-system registry with each source (read on 20 October 2026). A platform's documentation says what it can do; what one employer switches on can differ.
| Hiring Platform | What its documentation says | Source |
|---|---|---|
| Workday Recruiting | Autofill can fill contact details, work history and education from your resume, and applicants are advised to check what it filled. | Career sites (Workday) · State of Oregon applicant guide |
| Greenhouse | Accepts .pdf, .doc, .docx, .rtf and .txt files and cannot parse resumes larger than 2.5 MB. Tables, headers and footers, contact details in a header, footer or text box, columns, graphics and scanned files are listed causes of an unsuccessful parse. |
Supported file types · Unsuccessful resume parse |
| Lever | Resume parsing can fill your name, contact details, location and current company, and Lever advises checking them. Scanned files are a documented problem. | Understanding resume parsing |
| SAP SuccessFactors | Resume parsing can fill your contact details, current company and work history. SAP's knowledge base says it does not parse image files or scanned PDFs, and parsing covers 15 listed languages. | KBA 2081576 · KBA 2888759 |
| Oracle Taleo | A candidate's attachment is limited to 1 MB by default, and an employer can change it. Prescreening can include disqualification questions whose answers screen an application out automatically. | Attachment permissions and settings · Prescreening section |
What our 80-extraction layout study measured
To test how extraction software handles different resume layouts, we built ten fictional profiles, each in a single-column and a sidebar layout and in both PDF and DOCX — 40 files in total (not 40 people, and not 40 Guul templates). Three open-source libraries read the 20 PDFs and one read the 20 Word files: 80 extraction runs, with 11 expected fields per file (220 per library).
You can download the files, the raw text and the reproduction kit on our Resume Template Parse Tests page. Here is what the data showed:
| Extraction Engine | Input Format | Fields Extracted | Section Order Kept | What Happened |
|---|---|---|---|---|
PyMuPDF (fitz) |
220 / 220 |
20 / 20 files |
Kept every field and the declared section order in both layouts. | |
| Mammoth | DOCX | 220 / 220 |
20 / 20 files |
Reads the Word document's XML (word/document.xml); kept every field and the section order. |
Poppler (pdftotext, default settings) |
210 / 220 |
10 / 20 files |
Kept the order in all 10 single-column PDFs and mixed the two columns in all 10 sidebar PDFs. | |
Mozilla PDF.js |
178 / 220 |
20 / 20 files |
Kept the section order everywhere, but returned substituted characters for the phone number and date range in every PDF. |
What this means when you choose a template:
- Single-column layouts kept the declared section order in all 10 of their PDFs with every library we tested, Poppler included. Guul's single-column designs (
Crisp,Meridian) share that one-column structure, which is why we suggest them for large enterprise or legacy portals. The study itself used fictional layout files, not those templates. - Sidebar layouts kept the section order in all 10 PDFs with PyMuPDF and PDF.js, and in all 10 Word files with Mammoth, but Poppler mixed the two columns in every one. A sidebar design (
Onyx) works well when you want a compact one-page layout for human reviewers. Before submitting a sidebar PDF to an unfamiliar portal, open it in our Resume PDF Text Check or send a.docxcopy alongside it.
These are extraction libraries, not employer ATS products, and each file was read once: treat the results as a check on these files and settings, not a ranking of real hiring systems.
Your experience. A fresh canvas.
Try a different look. These are actual Guul template previews, with illustrative sample content.
A clean slate. Literally.
One column, distinct sections, and room for your experience to lead. A simple place to start.
0 multi-column reading notes across all 6 modelled hiring systems.
See full template parse table →Replace all sample details with your own experience.Explore the resume lookbook →
How Guul evaluates your resume against the Golden Standard
When you run your file through the Guul ATS Resume Checker, we do not invent a mystery grade. Every point comes from a rule you can inspect in our ATS Score Methodology, which publishes its weights and its version. Without a job description, a check scores three parts:
- File readability (35 points): whether the file can be read at all — real text rather than an image, a layout that survives being flattened into one reading order, and contact details in the body rather than a header (Rules 01 and 03).
- Structure (30 points): recognizable section headings, dates in formats that index, and entries with the fields a candidate record needs (Rules 02 and 04).
- Content quality (35 points): achievements rather than duty lists, bullets with enough substance to inform, and skills that appear in the experience that demonstrates them (Rules 05 and 06).
When you paste a job description, Job description match adds 40 points and the other three parts adjust to 25, 15 and 20, so the total stays 100.
If you want to see where your current resume stands right now, upload your PDF or Word file to the free ATS Resume Checker for a line-by-line breakdown, or read our ATS Guide hub and Explainable ATS reference to see how the score is explained.
Frequently asked questions
Does an ATS automatically reject resumes based on formatting?
Not by itself. Automatic rejections come from rules an employer sets, such as knockout questions about work authorization, a required license, or location. Formatting problems cause a different kind of harm. They can stop the system from filing your contact details, dates, or job titles into the fields recruiters search and filter. When a parser misses your dates or splices two columns together, your profile can show no tenure or garbled role titles before a person opens the file.
Is a PDF or a Word (.docx) file better for ATS parsing?
Neither is better everywhere. Greenhouse documents accepting .pdf, .doc, .docx, .rtf, and .txt files. For other systems, follow the file types listed on the employer's upload control. A text-based PDF keeps your layout the same on every screen. Neither format makes a risky layout safe: Greenhouse lists tables, columns, and text boxes among the causes of an unsuccessful parse, whether the file is a PDF or a Word document. Avoid scanned image PDFs, which contain no text unless the employer runs optical character recognition (OCR).
Can I use a two-column resume if I check the extracted text first?
Yes. A two-column or sidebar resume is not automatically unreadable. In our extraction study, PyMuPDF and PDF.js kept the section order in all 10 sidebar PDFs, but Poppler mixed the two columns in every one. Because extractors differ, test your exported file in the Resume PDF Text Check and check the autofilled fields on the employer's application form before submitting.
Why should I avoid skill rating bars, stars, and percentage gauges?
Graphic skill bars and star ratings contain no text for the rating itself, so a parser gets at most the skill name. Even for a human reviewer, "80% Python" has no shared meaning across companies. Listing your tools as plain text and naming them inside your work experience bullets gives both software and human reviewers verifiable proof.
What are the safest section headings to use on an ATS resume?
Use standard, literal section titles that parsers and recruiters both recognize: Summary (or Professional Summary), Work Experience (or Experience), Education, Skills (or Technical Skills), Projects, Certifications, and Languages. Creative labels such as Where I've Been or My Superpowers can cause the parser to leave your work history unindexed or merge it into a generic summary block.
