An engineer with eight years of strong experience emails her resume to a startup founder she really wants to work for. He opens it, glances, closes it, and moves to the next one in his inbox. Total time on her resume: 6 seconds. He doesn’t think he made a hiring decision in 6 seconds. The eye-tracking data says he did. The decision was binary: read further, or move on. She didn’t make the cut, not because she wasn’t qualified, but because her resume’s top third didn’t pass the scan.
What the eye-tracking data actually shows
Recruiters spend 6 to 7.4 seconds on the first pass of a resume. The original 2012 study by job-matching firm Ladders, Inc., recently re-summarized in their eye-tracking PDF, found that 80% of that scan was spent on six pieces of information: name, current title, previous title, current employer, previous employer, and education dates. A 2018 update raised the average to 7.4 seconds. A 2025 study by InterviewPal across 312 recruiters and 4,289 reviews found scans averaging 11.2 seconds, suggesting the number is moving up as more recruiters use AI-assisted tools but the underlying pattern is unchanged. The recruiter’s eye follows an F-pattern: top horizontal sweep, second horizontal sweep, then a vertical scan down the left margin. Anything outside that F gets minimal attention. The implication for resume design is the same now as it was in 2012: the top 5 inches of the page do almost all the work.
Why most resumes fail the 6-second test
The failure modes are predictable, and they cluster into four categories. We’ve seen these in thousands of resumes flowing through CurriculoATS.
- Visual clutter. Dense text, tight margins, multiple fonts, paragraph blocks instead of bullets. The eye gives up before parsing.
- Buried headline. Current title is on line 8 instead of line 1. The recruiter’s first 2 seconds get spent on contact info instead of the most relevant signal.
- No quantification. Bullets like “managed projects” instead of “led 4-engineer team to ship $2M ARR product in 7 months.” The eye scans, finds no number, moves on.
- Format penalties. Multi-column layouts, Canva templates, photos, sidebars. ATS parsers misread these 12-25% of the time, and human recruiters dislike them at roughly the same rate.
Of these four, quantification is the one that does the most damage with strong candidates. A senior engineer who has actually shipped real things often writes the most modest, vague resume in the stack. The 6-second test rewards specificity, not modesty.
The recruiter attention hierarchy: where the eye lands
Per the Ladders eye-tracking data, here’s where attention is actually distributed during the 6-second scan:
| Resume Element | % of Attention Time |
|---|---|
| Current job title and company | 27% |
| Previous job title and company | 16% |
| Education | 14% |
| Quantified achievements | 12% |
| Start and end dates | 10% |
| Skills section | 8% |
| Everything else | 13% |
The lesson, if you’re a candidate writing the resume: 70% of the recruiter’s attention lands on the top half of the first page, and that attention is hunting for title-company-dates-numbers. If your top half doesn’t contain those four things in scannable form, the rest of the resume might as well be blank.
The resume layout that passes the scan
The format that consistently passes both human and ATS scans in 2026 is unsurprisingly simple. Single column. Reverse-chronological. ATS-friendly font (Arial, Calibri, Garamond, Helvetica, Times New Roman) at 10-12pt body and 12-14pt headers, per Jobscan’s 2026 ATS template guide. Margins between 0.75 and 1 inch. No tables, no text boxes, no images, no sidebars. Submit as DOCX where the job board allows it (most reliable parsing); PDF only if the upload requires it.
The structure that consistently wins:
- Name + 1-line headline. “Senior Backend Engineer | 8 yrs distributed systems | Python, Go, AWS”
- Contact line. Email, phone, city, LinkedIn URL. One line. No address.
- Experience section, reverse chronological. Each role: company, title, dates, then 3-5 quantified bullets.
- Education. Degree, school, year. One line per credential.
- Skills. One line each for languages, frameworks, infra. No skill bars or rating systems.
That’s it. Resumes that look like this take 90 minutes to write and pass both the 6-second human scan and the parser. Resumes that look like a Canva infographic feel modern but get rejected at twice the rate.
What we learned at Amazon about ranking that applies to resume design
Before Curriculo, our founder Dev spent years on Amazon’s search and recommendations team. The most generalizable lesson from that work for any candidate writing a resume: any ranking system, human or machine, decides quickly using a small number of high-signal features. Optimizing for those features is not gaming the system; it’s communicating clearly. A resume that surfaces title, company, dates, and a quantified outcome in the top 5 inches is doing the recruiter’s job for them. A resume that hides those things behind narrative paragraphs is making the recruiter work harder than the next candidate’s resume, and the recruiter has 200 more to scan.
This is also why CurriculoATS Impact Scoring evaluates four signals (quantified achievements, experience relevance, career trajectory, skills alignment) and produces a written reasoning paragraph instead of a single keyword score. The reasoning paragraph mirrors the recruiter’s mental scan but makes it auditable. A candidate who passes the scan also passes the AI score, because both are looking for the same thing: clarity, relevance, and a numerical outcome.
What the 6-second test misses (and why founders care)
The 6-second scan is fast, biased toward formatting, and blind to context that often matters most for startup roles. A self-taught engineer who built and shipped a payments service used at scale frequently writes a flatter, less-buzzword resume than a candidate who has held three rotation-program roles at brand-name companies. The first candidate is more likely to be the right hire for a 30-person startup; the 6-second scan rarely surfaces them above the second. We see this pattern weekly inside CurriculoATS: the candidate the founder ends up hiring was ranked 24th by their old keyword ATS and 4th by Impact Scoring, and the gap was almost always vocabulary, not skills. The second blind spot is career arc. A 6-second scan catches current title and previous title, not the trajectory between them. A founder hiring their first engineering manager cares whether the candidate has done the 5-to-15 leadership scaling jump before, but that signal lives in the verbs of the bullets, not in the title-and-company line. Outcome-based ranking with a written reasoning paragraph is the structural fix: it surfaces the trajectory and the substance the human eye misses in 6 seconds, then hands the founder the paragraph to verify. The recruiter still owns the decision; the model just makes sure the right resumes are in the top ten.
The third blind spot is recency. The 6-second scan reads top-of-page first, which is recent experience for most candidates, but resumes formatted with a long career summary or a skills section above the experience block push the recent role down the page where the scan rarely lands. A 12-year veteran with three years of recent founder-stage work sometimes loses to a 5-year candidate whose recent role sits cleanly at line 3. Outcome-based ranking reads the whole document and weights recency by relevance, which is what a founder actually wants. The 6-second scan is a starting heuristic, not a hiring system; treating it as one is the source of the bias most founders complain about.
What founders should ask of their hiring stack
If you’re a founder receiving resumes, the 6-second test cuts both ways. You want a system that surfaces strong resumes you might miss in a 6-second human scan, especially when the resume itself is poorly formatted. Three things to look for in your ATS:
- Contextual parsing, not keyword matching. Strong candidates often write modestly. The system should infer skill from outcome, not just from keyword presence.
- Written reasoning, not a star rating. A 73 with no explanation is the same effort as reading the resume yourself.
- Auditable bias posture. Per NYC Local Law 144, automated employment decision tools require an annual bias audit. Ask any vendor for theirs.
Frequently asked questions
Is the 6-second resume scan still accurate in 2026?
The original Ladders study landed at 6 seconds; a 2018 update raised it to 7.4 seconds; a 2025 InterviewPal study across 312 recruiters reported 11.2 seconds. Whichever number you use, the conclusion is the same: the recruiter scan is fast and follows an F-pattern. Optimize for the top 5 inches.
Should I use a Canva resume template?
No. Canva templates often use complex layouts, multi-column designs, and image-based text that ATS parsers misread. The look-good-on-the-eye choice is the wrong one for the parser. Use a single-column DOCX with standard fonts.
What’s the most important section of a resume?
The top third of page one. That’s where the F-pattern scan lands. Name + 1-line headline + most recent role with quantified outcomes. Everything else is supporting evidence the recruiter only reads if the top third earned the second look.
How does an AI ATS see my resume differently than a human?
An AI ATS reads the whole resume but weights signals differently. It can compensate for buried information that a human would miss in a 6-second scan, but only if the parser can read your file. Multi-column layouts, embedded images, and unusual fonts cause parsing errors that defeat the AI’s advantage.
Should I tailor my resume to each job?
Yes, lightly. Match 3-5 keywords from the JD into your existing language without rewriting. Keep the structure and quantified bullets the same. Aggressive rewrites slow you down and rarely help; small keyword tuning is the highest-ROI tailoring move.
What if I have a non-linear career history?
Lead with outcomes, not roles. A candidate with a startup founder stint, a year of consulting, and a return to engineering shows up to a 6-second scanner as confusing and to an outcome-based scanner as differentiated. Use a 1-line headline that names the strongest current capability, then order the experience section by relevance, not strict chronology. Founders hiring their first 10 engineers often actively want non-linear histories.
What to do next
If you’re a candidate, time-box a 90-minute resume rewrite this week using the structure above. If you’re a founder, the 6-second test is also your test, and your tooling should help you not miss strong candidates whose resumes are imperfect. Walk through how AI resume screening works inside CurriculoATS and see what a written reasoning paragraph looks like in practice. The original eye-tracking research is at TheLadders’ published PDF, and the canonical 2026 ATS-friendly format guide is on Jobscan.