Resumes frequently fail to signal how good a fit you’ll be for a job. That's going to change.
Resumes are a poor proxy for a human being.
Whether on paper or LinkedIn, they may tell an employer about a job seeker’s experience and credentials, but they’re frustratingly silent about almost everything else. They have virtually nothing to say about a candidate’s personality, or their character, or their ability to persuade and communicate—all soft skills that employers consider essential ingredients for success.
“Resumes are terrible,” says Laszlo Bock, the former head of human resources at Google, where his team received 50,000 resumes a week. “It doesn’t capture the whole person. At best, they tell you what someone has done in the past and not what they’re capable of doing in the future.”
But even as a register of a job candidate’s professional history, a resume is horribly flawed. Its rigid format organizes a life’s experiences into bite-sized units designed for the consumption of hiring managers, or—increasingly—scanning software.
Resumes force job seekers to contort their work and life history into corporately acceptable versions of their actual selves, to better conform to the employer’s expectation of the ideal candidate. Unusual or idiosyncratic careers complicate resumes. Gaps between jobs need to be accounted for. Skills and abilities learned outside of formal work or education aren’t easily explained. Employers may say they’re looking for job seekers to distinguish themselves, but the resume requires them to shed their distinguishing characteristics.
Who—or what—is reading your resume
The fortunate candidate will have his or her resume read first by a human being; even then, the average hiring manager only spends six seconds (pdf) sizing up each candidate. More typically, a resume will first be read by an applicant tracking system, or ATS, software that’s programmed to search for combinations of keywords—and to spit out resumes that don’t meet the criteria. For many large employers, the resume exists only to identify reasons to disqualify candidates and thin the applicant pool.
Meanwhile, for all the valuable information that resumes exclude, they almost always include details that can actively work against job candidates. Signifiers like names, gender, home address, and educational background may all prejudice employers against candidates, either consciously or unconsciously.
As knowledge-economy employers seek out the workers whose creativity, drive, and leadership skills set them apart, the conventional resume is a relic. To correct the biggest weaknesses, progressive companies invest time and energy to “blind” resumes by obscuring details that could bias hiring managers.
To get it right, some corporations—and the startups and employment firms that serve them—are experimenting with new tools to determine a candidate’s fit. Assessments are being designed to identify the skills employers are looking for, and report them in easily digestible formats. And as pattern-recognition technology improves, software that matches candidates with jobs is becoming more sophisticated.
The resume of the near future will be a document with far more information—and information that is far more useful—than the ones we use now. Farther out, it may not be a resume at all, but rather a digital dossier, perhaps secured on the blockchain (paywall), and uploaded to a global job-pairing engine that is sorting you, and billions of other job seekers, against millions of openings to find the perfect match.
Even now, technology is upending the traditional relationship between candidates and employers, allowing companies to target prospects who may not be actively job-seeking. Algorithms are scouring social media to match jobs with user profiles, and proactively inviting members to apply, while startups are mining the discarded resumes languishing in the databases of bigger employers to unearth overlooked talent.
“We are moving to a place where the job seeker is seeking a job, but where a job is also seeking the job seeker,” says Steve Goodman, CEO of Restless Bandit, an HR software maker that specializes in so-called talent rediscovery. “In five to ten years, I see a future where the job finds the job seeker.”
The first resume
Leonardo da Vinci is credited in popular histories with writing the first resume. The actual evidence is a bit shaky, but in 1482, the Renaissance inventor wrote a letter to the Duke of Milan seeking employment, in which he detailed his accomplishments in developing instruments of war. A typical paragraph:
Where the operation of bombardment might fail, I would contrive catapults, mangonels, trabocchi, and other machines of marvellous efficacy and not in common use. And in short, according to the variety of cases, I can contrive various and endless means of offense and defense.
Until the 20th century, resumes weren’t separate documents, but like da Vinci’s, were included in job-application letters as a list of skills and accomplishments. By the 1920s, freestanding resumes became common enough to be described in business textbooks. First called “data sheets” or “personal profiles,” those early resumes included entries for education, experience, and skills, along with a list of references.
While the word resume (or résumé, if you’re fancy) has been used in English since the 18th century to describe a summary of past events, the first record of it being used to describe a personal profile, according to the Oxford English Dictionary, was an ad in a Lincoln, Nebraska, newspaper in 1926 that asked applicants to “Send resume of previous business connections in letter of application.” (In the U.K. and Ireland, the term curriculum vitae—latin for “course of life”—is used to refer to a resume, while in the US, a CV is a much longer document, often used by academics to list their history of publications).
Business correspondence manuals from the first half of the 20th century instructed candidates to use their cover letters to convey personality, with resumes reserved for factual details only. Accordingly, the human element vanished from resumes. They became less an instrument of persuasion than a bill of lading, or a list of ingredients.
Once resumes became instruments for marketing candidates as products, their writers began to refer to themselves in the third person, and adopted the bizarre convention of the subjectless sentence, where personal pronouns disappear.
This strange syntax produces sentences like “Maintain power control packages,” in place of more natural language like “I maintained the power control packages,” notes Lester Faigley, an emeritus professor of rhetoric and writing at the University of Texas, in his 1992 book In Fragments of Rationality: Postmodernity and the Subject of Composition.
“No one says ‘Maintain power control packages’ except those who wrote the institution’s official discourse and those who seek to identify with the institution in order to gain employment,” Faigley writes.
The language of the resume establishes and enforces the power relationship between the candidate and their would-be employer, he says, as “an initial gesture of subservience, like a dog presenting its neck.”
As employers began to expect resumes from every job candidate, a new cadre of professional resume writers began to form that helped further standardize the format. Often, they were former secretaries, skilled at writing and adept at formatting with typewriters, according to Mary Jo King, the president of the National Resume Writers Association, a nonprofit group with about 600 members.
The rise of the machines
The golden age of resume writing was in the 1980s and early 1990s, when personal computers allowed anyone to format and write their own resumes with ease. But the advent of the internet and email would soon mean job ads could be seen nationally, and resumes began flooding into large employers electronically. Software was developed to manage the flow by screening those new documents digitally, and choke down the number to a manageable trickle that could be read by humans.
Early ATS software was rife with bugs and couldn’t handle documents that arrived as a PDF, had fonts with serifs, or included graphic elements like charts, King says. They were engineered to look for keywords, and nothing else. The result was even more standardization—and candidates who couldn’t conform to the new expectations were left by the wayside.
“We asked the ATS developers, ‘What about all the great candidates you lose?’ and we were told it was irrelevant,” King says. “The idea was to derive one butt for one seat, and if they achieved that, it didn’t matter how many great candidates they lost.”
While the sophistication of ATS has improved in recent years, it’s also become much more affordable, so now the software is accessible to virtually every employer. Good luck to the candidate who doesn’t cram as many keywords into their resume as possible.
When resumes fail
The corporate world’s reliance on resumes—and its flawed assumptions of what belongs on them—means they can fail employers as easily as they fail candidates.
Claire McTaggart was a manager charged with recruiting for Monitor Deloitte’s strategy consulting practice in the Persian Gulf from 2012 to 2016. She helped identify the qualifications the firm used when choosing exceptional candidates to interview.
“We were saying, ‘We like people from Ivy League schools or from top schools, we like their GPA to be above a certain number, we like for them to have consulting or investment banking experience, we’d like them to have some charity work or not-for-profit experience,’’ McTaggart says.
“It’s basically going down the resume, and saying, ‘These are the things you have to have achieved in order to do well at our firm.’”
But after the new hires with the glittering resumes began at the company, McTaggart noticed they rarely were her best performers. The real stars came from unconventional backgrounds—one went to an obscure university in Jordan, another studied marine biology. They weren’t recruited out of college, but possessed the skills and attributes to excel at Deloitte. Usually they were hired because they were referred by another employee who sensed they might thrive at the firm.
Ivy League educations and experience in finance weren’t predictive at all, she says. “There was almost no correlation between our hiring parameters and the people who do really well.”
Ultimately, McTaggart realized, the resume “was just a way of targeting and filtering. You have the world of candidates—you have to be able to reduce that to a number. And you say, ‘If they’ve gone to Harvard, someone else has vetted them. They’ve gotten through this, so they must be good.’”
In her frustration, McTaggart recognized a business opportunity. Last July, she launched SquarePeg, a startup that tests the soft skills of job seekers and provides the results to employers, who pay for a curated slate of candidates. So far, 40 companies have signed up and 15,000 job seekers have taken the test and are looking for a job.
She’s betting employers will pay a premium to see small pools of targeted candidates, versus their current system of recruiting thousands, then using flawed systems to sift through them.
Squarepeg is just one of a number of startups looking to exploit the flaws of resumes and ATS, and provide a more sophisticated match between job seekers and employers. Gap Jumpers works to blind resumes by asking applicants to take a subject-specific test provided by employers, then offering the results stripped of names or identifying characteristics. CodeFights, a site which provides training for aspiring computer coders, connects tech employers with potential hires, depending on how the programmers perform on coding competitions on the site.
Another strategy is to introduce personality assessments earlier in the hiring process, to the widest part of the recruitment funnel, to prevent possible stars from slipping through the cracks in ATS, or from being dropped because of conscious or unconscious bias in the interview process.
Personality tests are not new in business, of course. But until recently they were primarily administered either late in the hiring process or as part of the onboarding of new employees. Now, at Goldman Sachs, for instance, they’re being piloted as part of the screening process.
Johnson & Johnson also is designing assessments to help it predict the performance of applicants for various jobs. The pharmaceutical giant is analyzing the traits of its most successful employees, and then building tests to measure those attributes in candidates, said Sjoerd Gehring, J&J’s global vice president for talent acquisition.
The goal is to eliminate needless rounds of interviews, while getting a more precise fix on the qualities in candidates that may actually help them succeed at the company. J&J hires hundreds of sales representatives annually, and has discovered optimism is a behavioral trait that’s predictive of success, Gehring says. So the company is building tests that can measure a candidate’s optimism.
Eventually, J&J hopes to have tests for a wide range of roles, even for ones as specialized as the researchers charged with discovering news drugs.
The testing at J&J extends to senior executives, not just entry-level recruits, Gehring says. “Often those people enjoy being tested,” he said. “If a company doesn’t assess you in a pretty substantiantal way, [candidates will say] ‘You guys aren’t serious about attracting the best talent.’ It’s been received with open arms.”
“Why do you need a resume?” asks Unilever
Two years ago, the North American division of Unilever—the consumer products giant—stopped asking for resumes for the approximately 150-200 positions it fills from college campuses annually. Instead, it’s relying on a mix of game-like assessments, automated video interviews, and in-person problem solving exercises to winnow down the field of 30,000 applicants.
“For the students, why do you need a resume?” says Katie Ambrose, a Unilever manager who helped design the company’s new hiring process. “For us, it’s more about about measuring potential than past.”
Applicants for Unilever’s internships and entry-level jobs now upload basic biographical data using LinkedIn. They then answer some questions about their employment eligibility (i.e., do they have permission to work in the US), which eliminates about half the field. If they’re eligible to work, they move on to an assessment developed by pymetrics, a company founded in 2013 by a pair of Harvard- and MIT-trained neuroscientists.
Pymetrics builds custom assessments for companies by testing at least 50 top performers at each employer to determine what set of traits lets them thrive at work.
Instead of asking questions designed to identify preferences, as SquarePeg does, pymetrics uses simple online games, like digitally inflating balloons until they pop, to measure traits like spontaneity, attentiveness, and flexibility. Unilever’s candidates play 12 games, which take about one to three minutes each.
Pymetrics CEO Frida Polli says the games are “culturally agnostic” because they don’t rely on language and that they’re harder to outwit than more traditional tests. The company performs “algorithm audits” to ensure the results are evenly balanced by gender, Polli adds. If the audits find a game is tilted in favor of men or women, the results are de-emphasized in the algorithm.
The games are designed so there are no wrong answers— a weakness in one characteristic, like impulsivity, can reveal strength in another, like efficiency—and pymetrics gives candidates who don’t meet the standards for one position the option to apply for others at the company, or even at other companies. The algorithm matches candidates to the opportunities where they’re most likely to succeed. The goal, Polli says, is to eliminate the “rinse and repeat” process of submitting near identical applications for dozens of jobs, and instead use data science to target the best match of job and employee.
One assessment to rule them all?
There are dozens of assessments in the market, ranging from Korn Ferry’s leadership assessment tool, designed for would-be executives, to Traitify’s simple, 90-second photo-based test aimed at first-time and hourly employees. Short tests risk being overly broad or inexact, while long ones can lead to frustration, and job candidates abandoning them before they’re done.
Given the variety, it may take years—or decades—for the hiring industry to coalesce around assessments that are universally accepted and recognized. And even then, J&J’s Gehring says, employers may still want to subject applicants to their own tests.
If there were a standardized assessment to emerge, there already is a natural home for it: LinkedIn, a platform used by more than 540 million professionals worldwide and virtually all major employers. Including a personality or soft-skills test as part of user profiles would be a powerful endorsement of their utility, and the developers of the tests are eager for LinkedIn to anoint their product as its assessment of choice.
LinkedIn, however, has been slow to embrace assessments. It sees them as just one of many possible solutions to the candidate-employer matching problem, says Ryan Roslansky, the global head of product at LinkedIn.
LinkedIn—and the broader recruitment industry—already can adequately sort candidates based on credentials and experience, he says. Determining whether someone will fit in and be happy, though, “is the Holy Grail of workforce planning, and there aren’t too many obvious answers about how to get there,” Roslansky says.
The LinkedIn effect
As it works now, LinkedIn both reinforces and undermines the authority of traditional resume. Its basic format and structure mimic chronological resumes, as does the emphasis on previous jobs and education. And its de facto requirement of a photo can open the door to employer bias.
But unlike resumes, LinkedIn offers candidates space to express themselves, to use personal pronouns, and to include recommendations from other users. As an increasing number of employers accept LinkedIn profiles in lieu of resumes, its importance as an arbiter of candidate information will only grow.
While LinkedIn may introduce more fields in its standard profile—to give candidates opportunities to provide more information, and algorithms more data points to sort them by— Roslansky says the company places more value on what others think of users than on what users think about themselves.
The company is exploring how it can match jobs and candidates by better understanding individuals through their connections with other LinkedIn members, he said.
“Connections as an indication of what someone knows, who they know, how they know them, what their strengths are, where they’ve worked—that’s ridiculously valuable data,” he says.
LinkedIn could build pattern-recognition software that combines such data with insights from companies into what combinations of employees are effective, he said.
“You can imagine scenarios where we ask recruiters, who have visibility into their organization or are trying to find someone new, about what types of people are working well together inside the company, where are there pockets where they’re seeing success,” Roslansky said.
Ultimately, LinkedIn could better match job seekers with employers based on what it learns about what sort of people thrive in different environments, he says. But there will be a lot of trial-and-error before it gets there, and at some point, LinkedIn may decide assessments are the best way to measure fit.
“We’re learning,” he said. “This is a hard problem, and no one has had this set of data before to try and solve this problem. I can’t point to someone else and say hey, let’s do what they did.”
Interview with a robot
Indeed, the world’s biggest job site, is approaching the problem from another direction.
Despite its reliance on resumes—it sees 2.6 million of them a month—it’s working on ways to make them obsolete, says Raj Mukherjee, who heads products at Indeed. One possibility is using language analysis, which it’s already doing for some employers.
For example, when a call center in the UK needed to hire representatives to respond to customers who speak seven different languages, Indeed developed an AI-enabled test where candidates were “interviewed” by a robot. The software evaluated their speech for syntax and grammar, and the company made offers based on the results, Mukherjee says.
In theory, Indeed could test for soft skills in similar way, by asking the right probing questions and performing a sentiment analysis on the answers.
But Indeed will embrace assessments of personality only if employers, and candidates, found the results useful, Mukherjee says. “I would hate to have a job seekers do something they never get value out of,” he said. “If you have another line on your resume that no one looks at, that’s not valuable.”
How Colossal Boy’s mom became president
At some point, resumes will disappear altogether, along with most of the traditional mechanisms of job placement, says Bock, the former Google HR chief.
“There’s 4 billion people on the planet who want to work,” he says. “What will eventually get built is a system that understands them as individuals. Then you won’t need a resume, and you won’t need to apply to anything.”
He predicts that jobs seekers and employers will be paired by a sophisticated third-party algorithm that has enough information about the characteristics of both workers and employers that it can play matchmaker.
Bock—now the CEO of Humu, a startup planning to apply science and machine learning to the workplace—is a fan of science fiction. As a kid, he collected the Legion of Super Heroes, a comic book about a team of super-powered teenagers in the 30th century. In that distant future, politicians didn’t run for office, but rather, the best candidates—including Marte Allon, the mother of the hero Colossal Boy—were identified from the world’s population by computer, and presented to the electorate for a vote.
It may not take a 1,000 years before we develop a similar system that can perfectly sort job seekers with jobs, but Bock recognizes that at present, there are substantial obstacles. More than just the programming challenge, it requires building the trust on the part of employers and would-be employees.
For the system to work, it would need an understanding of a company’s corporate culture, and how people actually function within its walls—not just what the company says about its culture. And employees and applicants would need to be comfortable handing over their personal data.
For-profit entities wouldn’t be trusted as stewards of such sensitive information. Nor would governments, Bock says, noting that in communist Romania, where he was born, “the government literally had dossiers on every single citizen.”
Ultimately, Bock says, the system should be maintained by a not-for-profit, non-governmental organization. “What I’m imagining, no human being should ever look inside this thing. You shouldn’t need to,” he says.
Perhaps the biggest challenge would be getting people who are happy in their jobs to enter the system, Bock says. “The very best candidates typically aren’t applying for jobs, because they’re already doing well.”
But a system in which every job holder was entered would produce the best outcomes. “That’s how Colossal Boy’s mom became president,” he said. “Finding the people who don’t want to be president, that how you get the best president.”