To review a CV in about ten seconds, look at four things in order: the current job title, the current company, the employment dates, and the education, then check them against the two or three non-negotiable requirements for the role. That is roughly what experienced recruiters already do. Eye-tracking research from Ladders found recruiters spend an average of 7.4 seconds on the initial scan of a CV, moving in an F-shaped pattern down the left and across the top. A fast review is not a careless one if it checks the right things in the right order and, crucially, never turns a ten-second scan into a final decision.
The part most advice gets wrong is treating speed as the goal. Speed is a constraint, not an objective. The objective is a consistent first pass: candidate number 180 assessed the same way as candidate number 2. Ten seconds per CV is fine for sorting the pile into “look closer” and “probably not.” It is not enough to reject someone for good, and the gap between those two uses is where fairness and good hires are won or lost.
The four things a fast scan actually checks
In the seconds you have, you are testing a small number of signals, and it helps to know which ones carry weight.
- Current title and company. The fastest read on level and domain. A senior title at a relevant company answers most of the “is this roughly the right person” question in a glance.
- Dates. Not to punish gaps, which are often irrelevant or unfairly read, but to understand the shape of a career: how long in role, trajectory, recency of the relevant experience.
- The one or two must-have skills. The specific, non-negotiable requirements for this role. If the job genuinely needs a professional qualification or a language, its presence or absence is a fast, fair filter.
- Overall legibility. A CV you can parse in seconds is doing its job. The Ladders work found simple layouts with clear headers and bulleted results held attention, while cluttered multi-column designs lost it.
Put as a quick reference, each signal has a job and a trap:
| Signal | What it tells you fast | What not to read into it |
|---|---|---|
| Current title and company | Level and domain | Prestige of the employer as a proxy for ability |
| Employment dates | Shape and recency of experience | A gap as a mark against the candidate |
| One or two must-have skills | A fair pass or fail on the role’s non-negotiables | Nice-to-haves treated as hard cliffs |
| Legibility of the layout | Whether you can assess it in seconds | Design polish as a stand-in for competence |
Notice what is not on that list: a deep read of every bullet, a judgement about “culture,” or anything you could not justify to the candidate. Those belong to a later, slower stage, not the ten-second scan.
A fast manual method versus a structured scorecard
The ten-second scan is the informal version. The moment more than one person reviews CVs, or the moment you need to defend a decision, you want the structured version. Here is the difference.
| Fast manual scan | Structured scorecard | |
|---|---|---|
| Speed | About 10 seconds | 30 to 60 seconds |
| Consistency | Fades over a long pile | Same criteria every time |
| Defensible later | Lives in your memory | Written record per candidate |
| Best for | Sorting into maybe or no | Deciding who to interview |
The scorecard is not slower for its own sake. It replaces an impression with a record: which criteria each candidate met and missed, scored the same way for everyone. That record is what lets you answer a hiring manager, or a candidate exercising their right to an explanation, without reconstructing a decision from memory. It also has to be kept lawfully: the Dutch supervisory authority treats deletion within four weeks of a procedure ending as the norm, so the scorecard is a record with a retention date, not a permanent file. It is the same principle behind writing your requirements as plain-language filters rather than a boolean string: readable criteria are reviewable criteria.
Where automation genuinely helps
Software can do the ten-second scan across a whole pile at once, and this is where the honest version of “let a tool do it” lives. A resume parser or screening tool extracts the same signals a recruiter looks for, title, dates, skills, education, and ranks candidates against your stated criteria, with the met-and-missed reasons attached; how those scanning apps actually work is worth understanding before you trust one. What it produces is a sorted pile and an explanation, not a verdict.
The distinction matters legally as well as ethically. A tool that ranks candidates and hands the ranked list to a person who decides is supporting the decision. A setup where software rejects candidates with no human ever seeing them crosses Article 22 of the GDPR, which restricts decisions based solely on automated processing that significantly affect someone. So the safe automation pattern is fixed: the tool does the fast first pass at scale, a human reviews the shortlist and the near-misses, and a human makes every reject and every hire. The tool never gets the last word, and it should not.
The limit of the fast scan
It is worth being blunt about what ten seconds cannot do, because the technique is easy to over-trust. A fast scan reliably tells you whether a candidate is roughly in the right territory. It cannot tell you whether they are actually good, whether their campaign ownership was real or borrowed, or whether the person behind a plainer CV would outperform the person behind a polished one. Those are exactly the false negatives a speed pass generates: strong candidates who present poorly on paper and get sorted into the wrong pile in seven seconds.
This is why the ten-second scan is a sorting instrument, not a judging one. Its entire job is to move two hundred CVs into a shortlist worth reading properly and a remainder worth a second glance, not to hand down verdicts. A recruiter who treats the fast scan as the decision has not saved time; they have moved the real evaluation to a moment when a hiring manager asks why a particular person was cut and there is no answer beyond an impression formed in seven seconds. The scan earns its speed only because a slower, fairer stage follows it.
The bias trap in fast reviews
Speed is exactly the condition under which unintended bias operates, because quick judgements lean on proxies. A name, a photo, a school, or a graduation year can nudge a ten-second read without the reviewer noticing, and there is strong evidence that identity signals change outcomes even when the underlying CV is identical.
Two safeguards keep a fast review honest. First, blind the first pass: hide names, photos, ages and schools so the ten-second scan runs on evidence, not identity. Second, use soft filters rather than hard cliffs, so a candidate who narrowly misses one requirement is flagged for a second look rather than dropped, which protects against the false negatives a fast scan produces. Both safeguards cost nothing in time and directly counter the specific failure mode of speed.
A worked example
Two hundred CVs for one marketing role. The non-negotiables are three years in B2B marketing and demonstrable campaign ownership. A pure ten-second manual pass would take over half an hour of unbroken attention, and by the last fifty the reviewer is not applying the same standard they started with.
The better sequence: a tool runs the fast scan across all two hundred at once, ranking them against the two must-haves and flagging near-misses, with names and photos hidden for that first pass. The reviewer now spends their real attention on a shortlist of perhaps twenty, plus the near-miss list, reading each properly, ten seconds becoming two minutes where it counts. Every advance and reject gets a one-line reason on a scorecard. The whole role is faster than the manual pass, more consistent, blind where it matters, and fully defensible, and not once did software decide anyone’s outcome. That is what “review a CV in ten seconds” looks like when it is done well.
How to review CVs fast without getting it wrong
Start by writing your two or three non-negotiables down before you open a single CV, because a fast scan is only fair if it is testing the same fixed criteria every time. Do the ten-second scan to sort into “look closer” and “probably not,” never to reject for good. Blind the first pass, use soft filters so near-misses survive, and put a one-line reason on record for anyone you advance or set aside. If the pile is large, let a tool do the fast scan at scale, but keep every actual decision with a person. Speed is safe when it sorts and dangerous when it decides.
Quick answers
How do you review a CV in 10 seconds? Scan four things in order: current title, current company, employment dates, and education, then check them against the two or three non-negotiable requirements for the role. Eye-tracking research shows recruiters average about 7.4 seconds on this first scan. Use it to sort into “look closer” or “probably not,” never to reject someone for good.
What do recruiters look for first on a resume? The current job title and company, which give the fastest read on level and domain, followed by dates and the one or two must-have skills for the role. Recruiters move in an F-shaped pattern down the left and across the top, so a simple layout with clear headers and bulleted results holds attention better than a cluttered multi-column design.
Can AI review a CV faster than a human? Software can run the fast scan across a whole pile at once and rank candidates against your criteria with reasons attached, which is genuinely faster than a person. What it must not do is make the decision: a rejection with no human involvement crosses GDPR Article 22. The safe pattern is a tool that sorts and explains while a person reviews the shortlist and decides.
Is a 10-second CV review fair? It can be, if it tests fixed, written criteria and only sorts rather than rejects. It becomes unfair when speed lets identity signals like a name, photo or school steer the read, or when a near-miss is dropped on a single missed requirement. Blinding the first pass and using soft filters are the two safeguards that keep a fast review honest.
What is the biggest mistake in fast CV screening? Letting the fast scan become the final decision. Ten seconds is enough to sort a pile but not to reject a person for good, and treating it as a verdict is how good candidates are lost and how automated-decision rules are breached. Keep the scan as a sort, and keep every reject and hire with a human who can explain it.