What Recruiters Actually Skim in the First Ten Seconds
They aren't reading your life story
Most screeners are trying to answer three questions: is this the right kind of job, are the dates continuous enough to interview, and is there one line that proves you did the work. They aren't grading your adjectives. If those three answers are buried under a paragraph about being a passionate go-getter, you lose the ten-second window.
I keep a printed stack of sample PDFs on the desk when we test templates. The files that survive a glance have a boring header, a current job with dates on the right, and a first bullet that names a tool or a volume. The files that fail have a photo, a two-line address, and a summary that could belong to anyone in the building.
A scan path you can design for
- Name and city — can I email this person.
- Current or latest title — does it match the req.
- Dates — any unexplained hole.
- First bullet of that job — is it work or fluff.
- Skills line — two or three tools from the posting.
Everything else is backup for the interview. Put the backup lower. Don't spend design energy on a skill-bar graphic. Nobody in a hurry trusts a self-scored “Excel: 90%.”
Where the ten seconds are actually spent
Eye-tracking studies keep finding the same shape: the first fixations land top-left (name area), sweep down the left margin like a zipper, and jump right once — usually to dates. Which tells you precisely what each zone has to deliver. The left column carries job titles and employers, so those words must be the honest, searchable ones a stranger expects ("Accounts Executive," not "Associate II, Revenue Pod"). The right jumps to dates, which is why date format consistency is a trust feature: "2022-Present" next to "Jun'21 Mar'22" reads as someone who didn't check their own document. The middle of the page only gets scanned after those two passes agree — which is the mechanism behind the oldest sad fact in recruiting: a great third bullet is a bullet nobody read. The formats article builds the whole page around this scan geometry; the practical takeaway here is that your layout isn't decoration, it's choreography for a very fast stranger.
What "right kind of job" means to a screener
Screeners don't evaluate careers; they match against a small internal dictionary of role shapes. "Is this the kind of person my manager described?" is answered in about three seconds by comparing your current title, your employer type, and one or two load-bearing nouns against the requisition. That's the entire first gate, and it explains a lot of rejections that felt unfair: a "Content Writer" applying to a "Product Marketing Associate" req fails the dictionary match even when the bullets are a real fit — and the honest parenthetical fix ("Content Writer — B2B SaaS positioning & launch collateral") is how you signal the match without lying about the title. The same logic decides what to put in the summary: name the target job family in its first five words, because the dictionary check happens before any sentence gets actually read.
Date continuity: what actually flags
Everyone worries about gaps; screeners barely register a clean gap. What registers is unexplained mess. Four jobs in three years with no pattern — the eyes stop and ask "what's the story here?" Two gaps that overlap an unlisted period — now it's a puzzle, and puzzles get solved by not-interviewing. And the one that kills quietly: dates that don't reconcile — a job ending "2023" in one line and "Jan 2022" in the skills list, an overlap nobody notices because the author knows the truth from memory. Fix all three cheaply: month-and-year everywhere, one format, and when the history is jumpy, one plain line explaining the pattern ("contracts through an agency; redeployed three times at the same ICU") in the summary or under the entries. An explained mess reads as life; an unexplained one reads as risk. The career-gap guide has the exact sentence templates.
What the first bullet of your current job must do
Here's where the ten seconds go if the first three gates pass: the very first bullet under your current role gets read — fully, out loud in the screener's head — and everything after it is optional. So that bullet carries your single best true match: the sentence where your work and this posting's problem overlap hardest. Not your most senior achievement, not your oldest responsibility — the most relevant one, reordered to the top without touching the truth. "Own the weekly revenue report the VP reads on Monday mornings" beats "assisted with reporting tasks" for a reporting-adjacent role even if you did both. The camera test from the editing walkthrough is how you write it: what would footage of you at your desk actually show? Name the object, the tool, and the volume, and let the second and third bullets be quieter. One strong first bullet per job is worth a page of balanced ones.
Second-pass reading: only after the first pass works
If a screener's ten seconds end positively, they come back later with something more like two minutes, and that's when the rest of the page finally gets its audience: the second job, the education line, projects, the skills list, a certification that matters. Which is the honest justification for keeping backup details low instead of cutting them — they're written for the reader who arrived, not the one who left. This two-stage model also settles most design arguments: make the second pass pleasant (clean spacing, consistent typography, no density) but never at the cost of the first. A layout that's beautiful in minute two and unclear in second one is a document that never gets seen. The practical corollary for the summary block: it's written for the ten-second reader but has to survive the two-minute one, which is exactly why it needs one number and zero decoration.
Agencies, in-house HR, and founders skim differently
The "recruiter" in recruiter advice is actually three different readers with different clocks. In-house HR screens against compliance-shaped requirements — degree, years, mandatory certifications — and their skim is checkbox-fast. Agency recruiters skim for sellability: current title, employer brand, numbers that can be pitched to a client in one sentence; they'll call someone with a great first bullet and a messy page. Founders and hiring managers skim for the job itself — they recognize work because they've done it, so the specificity that a checkbox screen misses ("fixed the reconciliation that breaks when a vendor changes GST rates") is exactly what lands with them. You can't write three resumes per application, but you can write one where the facts are concrete enough for all three readers to grab their own signal. That's the version of "audience-aware" that actually works — and it's the same reasoning behind the keyword method in our ATS article.
Run the test on yourself
Close the document, take a sheet of paper, and time your own skim: ten seconds, then write down the three facts you actually retained about a stranger's resume — yours, but read cold. (Printed helps; the phone works.) Then compare your three facts against what you'd want a screener to retain. Almost everyone finds the same two surprises: the thing they're proudest of didn't register at all, and something they considered minor dominated. Both are ordering problems, fixable in minutes. Repeat the self-test after every major template change, because layouts move your eye even when you know exactly what the page says. The version of this that hiring teams run on each other is called the corridor test — show your page to a colleague who doesn't work in your field, give them ten seconds, ask what I do. Their answer is your first bullet's performance review, delivered free.
ATS versus a human
The parser wants selectable text and normal headings. The human wants the same facts in a calm order. Those goals aren't at war. A MineResume Classic layout does both. Fancy columns look “premium” in a screenshot and messy after parsing.
If you only change one thing after reading this: rewrite the first bullet of your latest job so it starts with a verb and names the object you actually touched. Then export and try the ten-second test on a colleague who doesn't work in your field.