The information technology revolution is often portrayed as a job killer. ATMs eliminate the need for bank tellers, voice-recognition software has put many stenographers out of business, and payment-processing applications will reduce the need for checkout-counter workers.
But it’s also quite possible that number-crunching machines and algorithms could help reduce the unemployment rate, by tuning up the highly inefficient job-seeking and hiring process. That’s the bet Bright.com, a Silicon Valley start-up, is making.
As we gear up to dissect another monthly jobs report, analysts overlook a largely ignored problem. The Bureau of Labor Statistics reported that there were 3.7 million jobs open in the U.S. at the end of July, up a mere 0.2 percent from July 2011. But only 3.2 percent of those posts were filled in the month. Why aren’t available positions being filled?
Some experts claim that these “jobs open” numbers are lip service. Companies post jobs but refuse to fill them while they fret over Europe’s instability, market volatility, and the fiscal cliff. But there’s a larger defect at work, one that’s less the fault of the economy and more a byproduct of the way both job seekers and hiring companies use (or don’t use) technology.
Today’s job search is a goldmine of inefficiency. For starters, job seekers reach out to their swelled-to-bursting social network for leads. Then they comb through the millions of listings on Monster, CareerBuilder, and other sites, spending hours uploading millions of résumés (many of which aren’t right for the position), which employers then have to sort through. By the end, both sides have wasted an absurd amount of time and money. Job sites may hinder the process more than help, since applicants can each apply to hundreds of jobs. The creation of the multi-lane information superhighway has created a traffic problem.
Systems management companies are using big data to tackle the problem of congestion on the roads. Now serial entrepreneur Steve Goodman is using algorithms and cheap, powerful processing to develop a system to match employers and employees more efficiently. His company, Bright.com, is a machine-learning algorithm that aims to connect job seekers with the right jobs.
Launched in June 2012, Bright, based in San Francisco, works by creating a score for every would-be job applicant who visits the site. Users upload résumés and any other information they’re willing to share—location, Facebook page, etc.—and the algorithm delivers a score. It’s free for job seekers; the company makes its money by charging employers for its recruiting tools.
The trick—and it is quite a trick—is taking a host of ingredients, from the highly tangible (location, education, salary level) to the highly intangible (right level of experience, right background) and cooking up a single value that matches people with the jobs they not only want but can get.
The point of the algorithm is to have information that the individual user or human-resources executive can’t possibly access. Bright.com mines social- media contacts, and suggests jobs that come through people you already know. It pushes users toward companies that are more likely to hire from their college or grad school. It crunches numbers to determine a person’s preferred career path, even if she hasn’t thought it through herself, and suggests jobs she could be right for but may not have picked out of a search lineup. Salary is taken into account—applicants won’t be put up for jobs that are not in their desired range. (Users don’t have to share how much they make—the algorithm will guess.)
Once someone uploads their information and receives a Bright score, the site generates a list of jobs that present the closest numerical match. People can keep searching for more listings, but search results will be ranked according to the Bright score match. Users then decide which jobs they’d like to pursue.
Today’s job search is a goldmine of inefficiency.
To make the whole system work, Goodman hired dozens of human-resources professionals to evaluate piles of résumés and train the Bright score algorithm in how to think like the world’s fastest HR exec. He also hired a neuroscientist, Jacob Bollinger, as well as a former nuclear physicist, a geophysicist, and other data crunchers to build and test it.
To date, the site has listed more than 2.6 million jobs, mostly pulled from partnerships with Career Builder, Jobs Center, Beyond.com and around 20 other sites. The employers include a range of buzzy companies like Amazon and Zynga, plus Fortune 500 firms like Aetna, Nestle, and Wells Fargo. The biggest placement areas so far have been health care and telecom, as well as manufacturing—all industries that have helped make a dent in unemployment over the last few years. “Bright’s data-science team spends more time [focusing] on blue-collar and middle-office worker positions,” said Goodman. “Our sweet spot is people who make between $20,000 and $90,000—which makes up between 80 to 90 percent of America.”
For companies looking to hire, the upsides are clear: the same algorithms that spit out the right jobs for individuals likewise spit out the appropriate individuals for jobs. Bright allows hiring divisions to filter massive piles of résumés in seconds. What’s more, an HR professional, with seconds to spend perusing each résumé, may not realize that certain key skills or details indicate better qualification for a job—another problem that can be solved by technology.
Bright isn’t the only tech venture looking to remodel how we look for jobs—startups like Brewster aim to take our Facebook and Twitter friends and organize them into searchable networking categories that better enable us to find useful connections during a job search. And of course there’s LinkedIn, the pioneer, and reigning monolith, of online career development. But LinkedIn’s products focus primarily on leveraging your profile and connections to put yourself in potential employers’ line of sight, rather than handing the keys to a jobseeker. And while LinkedIn’s reach is huge—the company boasts over 175 million members, 62 percent of which are international—using all the site’s tools requires a level of savvy and knowledge that isn’t readily available to everyone.
While Bright couldn’t be more straightforward to use, its system is far from perfect—Goodman and Bollinger both acknowledge that it is still in the early stages. And while Bright has anecdotal reports of success from both job seekers and companies, there’s not enough data yet to hand our careers over to a jumble of code and say “go.” The good news is that, as a machine-learning algorithm, the more data the Bright score acquires, the more accurate it gets, and the more jobs are filled.
One thing we do know: simply moving along with our current hiring system isn’t an option—the Bureau of Labor Statistics notes that employers spend an average of $5,504 and two to six months per hire, while job seekers spend a median 19 months looking for their next gig. A lot of time, energy, and money are wasted as people attempt to find jobs. The economy would function a lot more effectively if jobs could find people instead.