Using AI to give job candidates a fighting chance
Personal experience with a seemingly “rigged” hiring system helped to inspire Vervoe co-founder Omer Molad to develop a system designed to facilitate truly talent-based hiring decisions. Launched last August, Vervoe has been used by more than 4,000 employers in 75 countries. Employee Benefit News recently spoke to Molad; edited highlights of that conversation follow.
Employee Benefit News: You seem to approach streamlining the hiring process from a somewhat different perspective than winnowing out the lesser-qualified candidates. Please explain.
Omer Molad, Vervoe: We’re about interviewing everyone, not about elimination, which is about 99.9% of how recruitment is done today. Other software tools are focused on better, faster screening, or how do you get people out of the funnel to save time. We do the opposite.
EBN: But the goal is still to save time, right?
Molad: Yes, but you can do both. Our approach is based on the belief that traditional screening methods do a disservice to both the candidate and the employer, because it’s done based on backgrounds, whether it’s their resume, LinkedIn profile, or any other static footprint. And we believe that to maximize your chances of getting the person who is going to be fantastic in the role, you actually have to see them in action performing tasks that you care about as an employer. The laws of physics make that very difficult, but we use technology to make that possible.
EBN: How does that work?
Molad: As soon as you apply for a job, you get a chance to prove your skills. At least at the first stage. You get a chance to do a task, you’re in the “room” now and if you are then eliminated, it’s not because you’re male or female or old or young, or you didn’t go to the right school, or you don’t have the grades or you are rich or poor. It’s because you didn’t perform as well as others in the tasks that the employer cares about.
EBN: But isn’t there a lot of subjectivity in the assessment of sophisticated skills?
Molad: Absolutely. We’re not prescriptive and we don’t believe that there is one way to hire. We don’t believe that there is a single best way to, for example, hire a graphic designer. We think it very much depends on the company, the role, the stage that the company is at, and we feel very much that a hiring journey needs to be tailored to that company’s circumstances. That’s why we’ve adopted a library model where expert assessment content can be deposited and employers can choose that and then build a really sort of tailored hiring journey to suit their unique situation.
EBN: At what point does the recruiter step in and take over from the system?
Molad: You set a set a threshold for the auto-grading part, so a candidate who gets above X percent automatically progresses to the next stage. And in about a month, we’re going live with an auto-ranking feature across the full gamut. We’re using machine learning technology to analyze, not just the substantive responses, but the manner in which candidates respond — how long it takes per question, do they log in and out multiple times or do it in one session, a whole range of factors that we take into account, also typos, mistakes, a lot of different things.
EBN: What does that additional kind of information tell you?
Molad: We use it in ranking the candidates we think are most likely to be progressed to the next stage of the hiring process. And the algorithm is learned — and we can do that based on text answers, video answers, mock-ups of PowerPoint, really anything. So the algorithms learn over time based on the actions that the employers take and who they progress and identify patterns and then they get smarter and smarter.
EBN: You have indicated that this system will limit the potential for hiring managers’ pre-conceived notions about who’s most qualified for a job to get in the way of hiring those who, objectively, are the most qualified. How’s that working out?
Molad: We don’t have a complete picture yet. But what we do know is that from all the interviews completed across our whole platform ever, more than 60% were completed by women. While I don’t know the percentage that end up being hired, it’s significant because we know that in many companies men are getting more opportunities, certainly in tech and certainly in senior roles. And so anecdotally, just based on that statistic, we’re doing something right and we have a very strong sense of why that is, and it’s because we’re putting performance ahead of background. Over time we hope to be able to collect a lot more data to close the loop around who’s getting hired and how are they performing after they’re hired and understanding the demographics.
EBN: Are employers concerned that job candidates might be uncomfortable with having a robot play such a vital role in the selection process?
Molad: Absolutely there are concerns about candidate adoption, like whether they will complete the process, especially for hard-to-fill roles where they are having to tap passive candidates. The concern is, “I don’t want to make them jump through more hoops.” But it’s actually fewer hoops because there’s a whole bunch of other steps that now become redundant; the candidates actually get to save a lot of time.
EBN: Tell me more about what you learn about a candidate by the way they actually go through the process.
Molad: It’s kind of like being observed through a one-way mirror. Let’s say we have two people do the same assessment and that they are roughly equal in terms of overall competence. But suppose one of them does the whole thing in one sitting, rushes through it, gets most of it right, a few mistakes, and hits submit. The second person takes three days, they log in and out 17 times, no typos, very cautious, check their answers. Same level of skill, very different approach. The first person, might be great at a startup company — they get stuff done, they hustle. The second person might be great working for a big auditor because they’re cautious, they’re diligent.
EBN: You have a library of assessments, but employers can create their own, too. Is that common?
Molad: Some of our customers have very strong views on how they like to hire and they are not looking for a third party to tell them what to ask, but they love our delivery method. One of the main use cases of the library, using someone else’s content, is when you are hiring for a role that you are not an expert in yourself.
EBN: How can the system evaluate criteria like a graphic artist’s approach to design, or a writing style?
Molad: Pretty much everything can be assessed on the platform. For writing, there are a lot of copyrighting, content marketing and email marketing tests on the platform and we have embedded the whole Google Apps suite. You can have a candidate open Google Docs and edit an email or edit an article. But if the employer wants to look at other data points that are not real time, for example, someone’s work portfolio or their resume, they can do that, and candidates can create a link to previous work. We’re not trying to prevent employers from doing anything, we’re just saying, “Here’s a way to see people in action, see them perform in real word scenarios.” And then you can back that up with a whole bunch of other reference points that are offline.
EBN: Did anything in your personal experience motivate you to develop this system?
Molad: I grew up in Tel Aviv, I went to the best high school in the country, did well and then became an officer in the military. But at the age of 22 I moved to Australia, and I hadn’t been to university; all I’d done is serve in the military. No one would give me an interview because I didn’t have a degree, had a funny name, and just didn’t fit into anybody’s concept of the sort of person they would want to hire. I found it frustrating going from being hot commodity in Israel to persona non grata. I felt like the game is rigged. And I think that stunned me. So then I put myself through law school and I kind of played the game and did things I needed to do to build up currency.
EBN: What about your partner?
Molad: Our co-founder, David Weinberg, had run a big technical team in Silicon Valley. The best performers always seemed to be the ones who didn’t go to Stanford. It was actually the ones who were unassuming, who didn’t have the resume, but were curious and applied themselves and worked hard.