You are looking at an advert for a job you would take tomorrow. Underneath it, seven bullet points. You have five of them cleanly, about half of the sixth, and the seventh is a piece of software you have never opened. Somewhere in the back of your head is the line everybody repeats: men apply when they meet sixty per cent of the requirements, women wait until they meet all of them. The moral is always the same, and it is always about you. Be braver. Apply anyway.
That number is worth following back to where it came from, because it turns out not to be a number.
Where the sixty per cent came from
The sentence appears in "A business case for women", by Georges Desvaux, Sandrine Devillard-Hoellinger and Mary C. Meaney, published in the McKinsey Quarterly in September 2008.1 "Internal research at HP showed that women apply for open jobs only if they think they meet 100 percent of the criteria listed, whereas men respond to the posting if they feel they meet 60 percent of the requirements." No citation is attached to it. It sits under a heading about the role of mentors, in a passage about encouraging female executives to seek out new positions, and is followed by an example about Lloyds TSB.1
In April 2014 Curt Rice wrote to the article's first author asking for the report behind the sentence. Desvaux replied: "This came from interviews with senior executives, which we conducted with many companies, including HP. I do not have access to the internal HP document, nor do I have the actual contact, as this was a series of confidential interviews."2 Rice's verdict was blunt. There is, he wrote, "no legitimate, evidence-based foundation for the claim", and "nothing that can be examined, reproduced, reviewed or cited".2 By then it had passed through Lean In and was on its way to being something everyone knew.3 The Behavioural Insights Team went back over it in November 2019 and reported the same thing: journalists had traced the claim to a speculative comment by a senior executive at Hewlett-Packard, with no quantitative data behind it.4
The article that put the figure in front of the largest audience is Tara Sophia Mohr's, published by Harvard Business Review on 25 August 2014, and it quietly undercuts the moral it is usually quoted for. Mohr surveyed over a thousand men and women, predominantly American professionals, asking why they had decided not to apply for a job when they did not meet all the qualifications. "I didn't think I could do the job well" came last. About 10 per cent of women and 12 per cent of men gave it as their main reason.3 It is a self-reported survey rather than a controlled experiment, and Mohr does not claim otherwise. It is also one answer among several she offered, and the rest of her results are not in the page available to us. But it is enough to make the interesting question a different one. Not whether you believe in yourself, but whether the list is enforced.
Whether the list is enforced is a question with evidence behind it.
The same job asks for more when more people are looking
Two papers by Alicia Sasser Modestino, Daniel Shoag and Joshua Ballance come at that belief from the employer's side, reading databases of online job postings collected in the United States by Burning Glass Technologies.
The first, published in the Review of Economics and Statistics in October 2020, covers 36.2 million postings from 2007, 2010 and 2012.5 The share of vacancies requiring a bachelor's degree or higher rose by more than ten percentage points between 2007 and 2010, then fell as the labour market recovered.5 A change in the mix of jobs being advertised could produce that on its own, so the authors go looking inside the categories. Within a single six-digit occupation code, a one percentage point rise in a state's unemployment rate came with a 0.6 percentage point rise in the share of employers requiring a bachelor's degree, and a 0.8 point rise in the share requiring four or more years of experience.5 It holds within firm and job-title pairs too: the same employer, advertising the same job title, asked for more. Using troop withdrawals as an outside shock to the supply of workers, they conclude that the rise in the number of unemployed workers accounts for 18 to 25 per cent of the increase in requirements between 2007 and 2010.5
The second paper runs the film backwards. Published in Labour Economics in 2016, and available in full as Federal Reserve Bank of Boston working paper 16-9, it uses 82.5 million postings from 2010 to 2014, the years the American labour market spent tightening. Requirements fell. A one percentage point reduction in the local unemployment rate went with roughly a 0.27 point reduction in the share of jobs requiring at least a bachelor's degree, and roughly a 0.23 point reduction in the share requiring five or more years of experience.6 The authors check the direction against the fracking boom, treating it as a shock to local labour supply in industries with no connection to natural gas. Their conclusion is that this reversed much of the increase the recession had produced.6
Put the two together and the reading is ours rather than theirs: part of what a requirement list asks for is tracking the number of people available to ask, rather than the job underneath it. How large that part is, the data cannot say, because a job posting never records what the work actually involved.
The limits are worth stating. Both papers are American. Both read what employers wrote in postings, not what they did in the hiring room. Both measure the two requirements a machine can reliably extract, education and years of experience, and say nothing about the rest of the list. The later window closes in 2014. We could not find an equivalent analysis of European postings.
Taking a requirement off the advert does not move the bar much
The opposite mistake is just as easy to make: treating a requirement's disappearance from an advert as evidence that the bar behind it has moved. In February 2024 the Burning Glass Institute and the Harvard Business School Project on Managing the Future of Work published a study of what actually happened at firms that removed degree requirements. Between 2014 and 2023 the annual number of roles from which employers dropped a degree requirement rose almost fourfold, having come down slightly from a peak in 2022.7 The adverts, in other words, changed a great deal.
The researchers then took 11,300 roles, each one a given occupation at a given employer, where they could observe a meaningful volume of hiring for at least a year either side of the change, and matched them against career histories for 65 million American workers.7 Firms that dropped a degree requirement increased the share of people they hired without a bachelor's degree by about 3.5 percentage points. That is a real change, and a small one. It also applies only to the 3.6 per cent of roles that dropped a requirement in that period, which leaves a net effect of 0.14 per cent — in the report's own phrase, "not even 1 in 700 hires last year".7 Nearly all of the movement came from 37 per cent of the firms studied, which went on to hire 18 per cent more workers without degrees into those roles.7 The others rewrote the advert and carried on hiring the same people.
So the two findings sit awkwardly together, and both are useful. The list is soft enough to inflate when the queue is long, and hard enough to survive being formally deleted. Neither of those can be read off the page you are looking at.
The part of the list that is a wall
Some requirements are not judgements at all. They are a yes-or-no question on the application form, wired to a rule.
Workable's documentation for employers puts it plainly: "For custom 'Yes/No' questions on the application form, you can choose to automatically disqualify candidates who answer 'No'." Several can be live at once, and a candidate is "disqualified if the answer is 'No' for at least one of the questions". The rejection is filed under the reason "Doesn't have required qualifications".8 The two uses the documentation itself gives are rejecting candidates who do not meet specific criteria, experience or qualifications, and rejecting applications from outside a region, its example question being "Are you based in the UK?".8 Recruitee describes the same mechanism in the same words: knockout questions "automatically disqualify and notify candidates who don't meet a job's requirements", and candidates who give the disqualifying answer "are automatically disqualified".9
This is the filter people picture reading their CV, and it is not reading their CV. It is reading an answer they typed themselves a minute earlier, usually about right to work, location, a licence, a language, or a minimum number of years. What you can see before you commit any effort is the question. Whether a rule is wired to it is a setting the employer chooses, and Workable's page says exactly that: an employer can choose to disqualify on a given answer. So a yes-or-no question on a form is not proof that a requirement is enforced automatically. It is the place where one can be, and the only warning you are going to get. Open the form first.
And the part where "essential" means essential
Some employers publish the rule they shortlist by, and where they do it can be a stricter one than anything above. NHS Lanarkshire's account of its recruitment process describes shortlisting as beginning by identifying applicants who have "demonstrated that they have met the minimum criteria in the person specification", with the desirable criteria then used to separate the people who cleared that bar.10
Two things follow where an employer says that. The essential column is a threshold rather than a preference, so the question in the title has a dull answer: meet them all. And the evidence for each item has to be in the application itself, because the application is the document being scored. A panel working to that procedure is not in a position to be impressed by the four criteria you did evidence and wave through the one you did not, which makes the evidencing the better place to spend your time than the deciding.
This is one NHS board's published procedure, and the essential-and-desirable person specification is not unique to it. But how strictly any given employer applies its own format is a fact about that employer, published or not. Where it is published, read it. Where it is not, you are back to the private-sector case below.
So, apply or not
Open the application form before you decide. Its yes-or-no questions take a minute to read, and they are where a requirement can be enforced without anyone thinking about you. You cannot see from the outside whether the employer switched that on, so treat each one as if they did: if a rule is running, no amount of care later in the form changes what your answer did. Answer them honestly anyway: a false answer that carries you past a rule surfaces at the offer stage, which is a worse place to lose the job.
Sort the rest into binary and graded. Right to work, professional registration, a licence, a clearance, a working language, a location you can be in: these come out as yes or no whether a rule enforces them or a person does. A number of years, a degree, a named tool, "experience in a fast-paced environment": these are graded, because somebody weighs them. Only two of them have been counted. The studies above measured education and years of experience, and those are the two shown moving with the labour market rather than with the job. What a named tool or a line about pace is doing on the list is not something anyone has measured.
Read the employer's own rule when there is a person specification. Employers using that format often publish how they shortlist against it, as NHS Lanarkshire does, and where they do, the essential column is the rule you are being read against rather than a guess about how somebody will weigh your CV. Find that page before deciding what an unmet essential item costs you: it is one employer's procedure, not a law, and it is the only version of the answer that applies to the job in front of you.
Everywhere else, read the count as weak evidence rather than as a verdict. Be careful about how much weight it can carry, in either direction. What the posting studies support is narrow: a list written when a lot of people were looking for work is partly a description of how many applications the employer expected, and the same employer advertising the same job title asked for less when applicants were scarce. None of them followed an under-qualified applicant into the hiring room, so none of them can tell you what missing two items does to your chances. And a requirement quietly removed from an advert is not an invitation either, because most of the firms that removed one carried on hiring the same people. What is left is a list that describes the employer's expectations more reliably than it describes your odds, which is an argument for deciding on the binary items and the strength of the fit rather than on the score out of seven.
And do not conclude that applying is free. Every application costs time that cannot go into another one, and this article has not shown you anything about how many to send. The reason to be careful about the seventh bullet point is not that it will disqualify you. It is that reading it properly is how you decide whether this is one of the applications worth doing well.
If the effort you are trying to protect is the effort of opening forty adverts to find the six worth applying for, that is a sorting problem before it is a confidence problem. JobCraftly will do the searching and the reading, and show you the roles that genuinely match what you asked for, with a note on why each one fits. What it cannot tell you is which of the seven bullet points this particular employer will enforce. Nothing outside the company can.
The list was written before you turned up. Some of it describes the job. A good deal of it describes how many people they were expecting to hear from.
