The Recruiting Game
Can gaming and testing combined with big-data analytics transform the recruiting space? Some think so. Others aren't so sure.
By Tom Starner
Effective recruiting remains one of the crown jewels of the human resource experience. And even with all the tech-driven bells and whistles applied to the function over the past 10 to 15 years, HR and recruiting leaders are still looking for better approaches to landing top talent.
Along those lines, an emerging wave of companies is making some strong claims about their abilities to make recruiting and screening more effective. Some have applicants playing virtual games; others use a data-rich form of pre-hire assessment testing. Some offer a combination of both games and testing. While the tools and companies may differ, they all rely on metrics, analytics and algorithms fueled by big data. The claim is that, by using these new tools, employers will -- at long last -- put the best possible people in their organizations' seats.
Companies such as Knack >, Evolv, CubeConnect, Prophesy Sciences, Smarterer and a host of others have been gaining notoriety in the mainstream business press by saying their products and services can get employers as close to recruiting perfection as possible. To do that, they also share a "Moneyball-like" approach, the metrics-driven methodology used successfully by Oakland Athletics General Manager Billy Beane and popularized in the film Moneyball, starring Brad Pitt as Beane.
< Knack > founder and CEO Guy Halfteck, an Israeli entrepreneur, is not shy about making predictions for this latest iteration of recruiting's silver bullet. In a way, he speaks for all the innovators behind this movement.
"The current recruiting continuum eventually will be obsolete," says Halfteck, a former attorney who built < Knack > after undergoing a frustrating career-change experience. "We are not offering a new incremental innovation on an existing assessment tool. We offer a better tool, a more strategic tool for companies to build their next-generation workforces."
< Knack >, in Palo Alto, Calif., uses video games as its entry for job candidates. One game, Wasabi Waiter, takes 10 minutes to play and casts players as waiters in a sushi restaurant. They must manage customers, dole out advice and serve to the best of their ability. In every < Knack > game, each decision is recorded and transformed into data by special sensors that enable algorithms to process player behavior. The game sessions, Halfteck says, allow < Knack > to deliver accurate assessments of traits such as creativity, persistence/diligence and other characteristics that are hard to discern from a resume, college transcript or interview.
As a gauge of its effectiveness, Halfteck points to results of pilot projects testing < Knack >'s technology, primarily a trial < Knack > ran with a unit of oil giant Shell that exclusively focused on technological innovation. Shell wanted to find more efficient criteria in choosing new employees for this division and asked < Knack > to put about 100 potential job candidates to the test with Wasabi Waiter and Balloon Brigade. < Knack > used its games to generate a list of candidates with the greatest potential from a pool Shell provided. As it turned out, Halfteck says, Shell reports that the 10 percent chosen actually did deliver the most innovative ideas, according to Shell.
Nanette McWhertor, vice president of operations and people development at Stacked, a Thousand Oaks, Calif.-based restaurant group with just over 400 employees, used < Knack > games in a pilot program. McWhertor says Stacked was looking for a new approach to testing for both managers and hourly team members.
"In both cases, we were looking more for specific behaviors and personalities," she says. "Those are hard to draw out in an interview and that's exactly what this 10-minute < Knack > experience does."
McWhertor had Stacked's most experienced, successful people take the < Knack > tests as part of the pilot. The results were about 95 percent accurate, she says. Stacked has committed to using < Knack > in its recruiting/hiring process for one year.
"It's very innovative, though I am not sure it's going to revolutionize the hiring industry," she says. "It does give hiring managers more insight into the person ahead of time, rather than waiting for those traits to blossom on the job."
At two-year-old ConnectCubed, based in New York, CEO Michael Tanenbaum says his company uses big data to tailor its online games to evaluate candidates for any position. Using personality surveys along with simple spatial-reasoning, trivia or memory games, ConnectCubed tests a company's longtime stars to develop ideal behavioral profiles for each job that needs filling. ConnectCubed games reveal information about knowledge and talent and then ranks game players. Employers purchase these rankings from ConnectCubed and weave this information into their recruiting and hiring processes.
Tanenbaum says CHROs and other decision makers who have seen ConnectCubed's potential have been impressed, but for the trend to go mainstream, employers need to be more "data trusting," which he admits is a challenge.
"Many are scared of data," he says, adding that he is currently seeing a move toward a combination of roles in HR within large companies. "People used to succeed as HR leaders because they were 'people people.' And that was fine," he says. "But now that senior management is seeing the benefits of a more rigorous data approach in many areas, HR leaders need an expanded skill set that embraces big data and metrics in recruiting and hiring. That may be scary, but it does not have to be scary."
Tanenbaum says the goal is to transition away from traditional ways of hiring -- typically based on instinct/gut feelings, interviews, resumes, etc. -- to a more data-centric approach, which he says is working for those who try it.
"No one is abandoning the old way, but all the research suggests that incorporating tests and games in the process delivers a higher predictor of success," he says.
Another competitor in the space, Evolv, uses surveys -- not games -- to drive its metrics approach to recruiting and hiring. Carl Tsukahara, executive vice president of marketing and products at Evolv, says his company uses surveys and big data to evaluate candidates, mainly hourly workers, and the results to date have been excellent. While < Knack and ConnectCubed are relatively new companies, Evolv has been fine-tuning its big data approach since 2007.
For more than seven years, San Francisco-based Evolv has collected 500 million data points by tracking the results of all its surveys and candidates' real-life employment histories. For example, it measures how long an employee has stayed with the company, job performance metrics (i.e., customer satisfaction), even the distance an employee lives from the workplace. Evolv's predictive-analytics recruiting product, called Evolv Selection, then translates new applicants' results into a "traffic light" system for hiring managers, with green meaning an applicant is "high potential."
For client Harte-Hanks, call-center workers who were selected by Evolv missed about 29 percent fewer hours of work in their first six months. They also handled calls 15 percent faster than those hired before the company began using the software. At Xerox, which did a half-year trial with Evolv, attrition was cut by 20 percent. Due to the success of the pilot, Xerox expanded the Evolv program across a population of 20,000 call-center and other back-office positions.
Tsukahara says Evolv's data often debunks common recruiting assumptions. For instance, Evolv's analysts have found that a history of job-hopping or unemployment is not a good predictor of how long a person will stay on a job. Tsukahara adds that Evolv focuses on evaluating front-line workers right now because performance data for employees engaged in higher-level tasks "is not there yet."
"Right now, our customers tend to be employers [that] already use data-driven processes, such as large credit-card companies," he says. "They understand the importance of data to the business outcomes."
Jay Goldman, a managing director with Klick Health and co-author of The Decoded Company: Know Your People Better than You Know Your Customers, says these emerging companies, and their big-data-based gaming solutions, are interesting because they focus on the difference between ambient data (data gained through game play and tests) and self-reported data. A resume, interview answer or even a reference are examples of the latter, which Goldman says are very often weak ways to gauge a job candidate.
"Self-reported data is inherently biased, intentionally or unintentionally" he says. "It's filtered and not very reliable, and that's why interviewing doesn't really work. Some of these new companies have hit on the principles that ambient data is more valuable. Data from a well-designed game, for example, can create a much more accurate view as a person than a self-reported view."
John Sumser, the founder, principal author and editor-in-chief of the HRExaminer online magazine and a noted HR tech guru, says when he thinks about gaming and big data applied to talent acquisition and recruiting, he envisions those early, crazy film clips of would-be flight pioneers unintentionally creating comedy while trying to get all sorts of oddball machines up in the air.
"I am absolutely persuaded that the future impact of big data on recruiting will be stunning," Sumser says. "But the scene today is like what those magnificent men in their flying machines were to today's airline industry. Like those early air experiments, a lot of it will just not fly."
Mainly, he says, these emerging players need to show the cause-and-effect connection between their products and recruiting success.
"Part of what will happen with this next generation of data-intense technology is organizations will have no way to understand what to do with it and if it really works," he says. "They need to test what works and what doesn't work."
Gerry Crispin, an Internet recruiting pioneer and co-founder of CareerXroads, a consulting practice based in Kendall Park, N.J., is not sold on the idea of gaming as the latest recruiting Holy Grail.
Like Sumser, he also has a sense that, while the applications are interesting, they are currently lack cause-and-effect proof, the "scientific method," Crispin says.
"Whatever data you collect through games, for example, it has to predict something downstream and then that prediction has to be tested," Crispin says. "If I am saying, 'I can get you better candidates or employees,' someone has got to prove that to me -- showing me exactly how that collected data found a better employee.
"That is often what's missing and without it, it's just entertainment," Crispin says.
For these tools to work, he says, they have to be applied to a specific job, not on some observed correlation based on a generic job and a generic population that is supposedly "similar" to the company's open position. "Right now," he says, "I don't think it will work for most of the games and simulations that are bubbling up."
Crispin points to Shaker Consulting Group in Cleveland as an example of a long-time supplier of accurate big data used to improve recruiting and hiring.
Shaker Consulting offers employers a tool called Virtual Job Tryout, which creates custom simulations for pre-employment testing and employee selection. According to Shaker President Brian Stern, the further the evaluation experience is from the content of the job, the greater the "leap of faith" that it will intentionally (versus accidently) measure job-relevant characteristics. On the flip side, he says, the closer the evaluation experience mimics the specific performance demands of the job, the more likely the evaluation will be relevant and meaningful to both candidate and recruiter.
"There is a job-centric approach or a test-centric approach," Stern says. "We start with the job, look at the job, the key performance drivers and then drive the assessment. It's not an exercise of hope.
"Everyone is looking for a silver bullet, but this isn't anything new," he adds. "There has been so-called big data for insurance, financial and marketing for 50 years."
A senior director of talent-management analytics and solutions at a major services company, and a Shaker Consulting client who asked to remain anonymous, finds using video games as part of the assessment process an interesting proposition that may become part of the future landscape. But his concern isn't so much around the ability to create a fun game that can predict job success, but rather around the potential adverse impact that a video game-based assessment may have overall.
For example, he says, just being comfortable with video games may relate to performance in the game. So groups that play video games more frequently (e.g., younger males) may have an advantage over other groups simply because of their video-game experience, not because they have more of the skills needed for success on the job. Unless, of course, the job involves playing video games.
"Would this build in a bias against certain groups? Maybe," he says. "The issue of test fairness is a really critical issue, and moving video games into the recruiting and assessment space will likely bring that issue to the forefront."