The six-second CV scan: what the study actually measured

No, and the study the figure comes from never claimed they did. The “six seconds” measure how long an initial decision took, keep or reject, made by thirty recruiters observed in 2012 by a company that sold CV rewriting. The trade turned it into a reading time, then a rule of thumb, then a sales argument, and none of those three versions appears in the original document.

The case deserves a closer look because it is the mirror image of the one we covered with the counter-offer statistic. There, the figure had no source at all. Here the source exists, anyone can open it, and it says roughly what people make it say. That “roughly” is exactly where the cost lies.

Where does the six-second figure come from?

From a five-page document published in 2012 by TheLadders, an American job site, under the title “Keeping an eye on recruiter behavior”, and it takes less time to read than most people spend quoting it.

The method fits in one sentence: thirty professional recruiters were followed with eye tracking for ten weeks while they reviewed CVs and online profiles. The equipment records where the gaze lands and how long it stays there. The study set itself three questions, one of which was explicitly to contradict what recruiters said about themselves, namely that they spent four to five minutes per CV.

The rest of the document is more revealing than the number. It compares CVs before and after a professional rewrite, scoring the rewritten ones 6.2 out of 7 for usability against 3.9 for the originals, then compares TheLadders profiles with LinkedIn ones, which it finds 55% harder to read, and closes with three recommendations: have your CV rewritten by a professional, look after the visual hierarchy, and follow the style of TheLadders profiles. The last line is a link to the company’s CV rewriting service.

None of this makes the observation false. It makes it market research, run on a small sample by a party with a direct interest in the conclusion, and never replicated by an independent third party.

What did the six seconds actually measure?

The length of the keep-or-reject decision, and the study says so itself: recruiters spent about six seconds on their initial fit or no-fit decision. That is not the time spent on a CV. It is the time spent before deciding whether the CV deserved more.

The difference changes everything. A CV rejected in six seconds did get six seconds. A CV kept after six seconds is then read again, more slowly, compared, sometimes printed, and none of that second read enters the measurement. Saying that “a recruiter reads your CV in six seconds” is like timing how long a bookseller spends on a back cover and concluding that they read novels in six seconds.

The study’s other finding is quoted far less, and it is the more useful one. During that first look, almost 80% of the time went to six fields: name, current title and company, previous title and company, the dates of both roles, and education. Beyond those, recruiters skimmed the text for keywords, and the detail of a candidate’s achievements, in the document’s own words, became filler with no effect on the initial decision.

What happened between 2012 and today?

The figure was updated once, and the update says more about how a statistic is made than about recruiters.

In 2018, Ladders published a new study, run in two stages, that gave 7.4 seconds. The company explained the gap by the context of 2012: coming out of the recession, the flood of applicants forced recruiters to move faster. The explanation is plausible, but it concedes that the number depends on the volume of the moment rather than on some constant of the job. And as a recruiting professional pointed out in ERE in 2020, the 2018 report gives neither the number of recruiters observed, nor their experience, nor the types of role, nor the length of the CVs reviewed.

Meanwhile the 2012 figure carried on alone. It got rounded, it lost its method, then its nature as a first screen, and today it circulates in career advice, recruiter training and the pitch decks of automated screening tools with the confidence of a law of physics.

It has an exact cousin, born the same year. “75% of CVs are rejected by the ATS before a human sees them” goes back to a sales pitch by Preptel, a CV optimisation company that closed in 2013 and never published a method or a sample. Two figures from 2012, both sold by someone who was selling the cure, and both still quoted in 2026.

Can a sourced figure be wrong?

Yes, and it is the hardest kind to spot, because it survives the usual check.

Faced with a statistic that has no source, the reflex is simple: look for the study, fail to find it, drop the figure. With the six seconds, that reflex fails, since the study exists and does contain the number six. What is wrong is the slide between what it measured and what it is made to say, and that slide only shows if you read the document to the end, which almost nobody does.

Hence a reading rule to add to the one about sources. Before repeating a figure, check three things: who measured it and what they were selling, on how many cases, and above all what exact quantity it measured. The third question disposes of more quotations than the other two combined, and it is increasingly asked by the person you are talking to, who can check in a minute with an assistant what you have just told them.

Why does the first screen penalise IT services consultants?

Because the six fields that catch the eye are precisely the ones that say nothing about a consultant.

Take a developer who has spent eight years at an IT services firm. Their current title is “consultant”, their employer is the firm, their dates cover eight unbroken years, and their previous role is often of the same kind. Everything that qualifies them, the clients, the projects, the technologies, the size of the teams, the nature of their responsibilities, sits in the detail of their assignments, which is to say in the part the study describes as filler. A first screen built on titles and dates reads that profile as a career standing still.

That is reasoning rather than measurement, and we present it as such. But it matches what every account manager in the trade already knows: a consultant’s CV is a poor sales document, which is why the industry invented the skills dossier, organised by assignment rather than by employer. Cobalt has gone through the mistakes in those dossiers that lose assignments, and most of them come down to the same thing: putting the assignment back in front.

The same bias hits career changers, careers with a gap, and unusual job titles. Those are the profiles the trade says it wants to look at more closely, and they are the ones a six-field screen turns away fastest.

What if the six seconds have become true?

That is the serious objection, and it deserves better than a shrug.

Volume has exploded since 2012. In June 2025, LinkedIn was handling around 11,000 applications a minute, 45% more than a year earlier, and a good share of that growth comes from applications written by generative tools that pick up every keyword in the ad. Faced with several hundred applications for one role, a quick first look is less a matter of laziness than the only way to reach the bottom of the pile. We have described elsewhere how this race between candidates and recruiters, both armed with AI, degrades the channel for everyone.

So the concession is real: a quick screen is sometimes necessary, and the 2018 study itself confirms that its length follows volume. What does not follow is the conclusion usually drawn from it. If the first look is short out of necessity, it should at least land on the right fields, and leave a trace of what it turned away and why, rather than vanishing along with the application.

What this changes for a recruitment agency or an IT services firm

The length of the first look is the wrong indicator; what it leaves behind is the right one.

A decision that turns someone away closes a door, and it should be explainable in one sentence: what was read, what was missing, what could not be checked. It is the same requirement that makes us prefer a ranking that explains itself to a match score, and it holds whether the screen is done by a person, by a tool, or by both.

The simplest test needs no software. Take twenty or so applications rejected at first glance last month and have someone else read them again, giving them three minutes per file and asking them to read the assignments before the titles. Count how many they would have called back. That number is yours, it owes nothing to a 2012 study, and it tells you better than any statistic what your first screen is costing you.

This is also where an agent makes a difference, provided it is put in the right place. Reading a whole CV, assignments included, takes Balt no longer than reading six fields, and what it returns is a sentence that can be argued with rather than a verdict. The decision to turn someone away stays with the person recruiting, because it closes a door on someone with a name.

That leaves the question of what happens to the candidates who pass the first screen. That is where time is really lost, in days of waiting rather than seconds of reading, and silence there costs more than a rejection.