Robert Park is a founder of SocialCheck.me, which wants to transform the way employers hire people. You know that little footnote to your resume, where you list a few references—mostly people you’re sure would never say a word against you? SocialCheck.me wants to crack that section wide open, putting mechanisms in place for employers to cast a broader net to more references who will give more, and more truthful, information about candidates. Currently in beta, SocialCheck.me will be expanding beyond its five users in the next month, and intends to be out of beta by the end of this year.
FAST COMPANY: How does SocialCheck.me work, step by step?
ROBERT PARK: The employer will put up a job posting and receive a bunch of resumes or inquiries. The employer notifies a candidate: We’re interested in you, we’d like to do a reference check regarding your work habits and past history, via your former colleagues and friends—could you please send this survey to them? They send candidates a link, and via the link they can log into Facebook or LinkedIn, choose specific people to respond, and those people will receive a notification in LinkedIn or Facebook from the candidate saying, “Could you please help me get this job by providing reference information for me?” Those people then click to a customized and detailed survey, and enter the information, and click submit.
Anonymity is key to the process.
It’s all anonymized, so that reference contacts feel they have a safe environment to make as much information as possible to the employer. Even candidates themselves are not aware what questions are on the survey. It’s a difficult problem we’re trying to solve: how to create the safest environment to get the most reliable data. We definitely believe we’re headed in the right direction.
Can respondents elect to show their name?
Currently, by design, there’s no way to do that.
Can employers at least see whether this answer to this question and that answer to that question came from the same person?
No, it’s all aggregated together and presented as one big data set. That is again by design—to try to keep individual respondents from ever having the chance to be called out. We don’t want employers to look at the data and do educated guessing: “This data must have come from that friend—we should contact that person directly.”
This is a funny optimization problem—you want to maximize truthful information while also maximizing privacy. What other systems did you study? I thought of Catholic confession, for some reason.
That’s an interesting analogy there. The part that makes that so safe is it’s something where you have a lot of confidence things will not come out. Our original inspiration came from 360 reviews, which enabled us to provide feedback or receive feedback to figure out how to improve ourselves. Hiring here is really only the first goal. We’re talking about maximizing trust in anything where there’s an asymmetric aspect of information. That could go for buying and selling on eBay, or dating—there’s a lot of potential here. The overall vision here is to create a trust infrastructure, and it’s in a very experimental stage.
I think even if I were anonymous, I’d feel bad saying anything other than nice things about someone applying for a job.
There have been times in my past where people have asked me to act as a reference, and I’m like, “Yeah, sure, okay.” But you’re not sure what you’re going to be saying. When the day comes and you’re asked the questions, you’re not sure what to say. The whole process ends up becoming very bland and not helpful for employers. Our core premise for our thesis is: Because references now are not anonymous, they do not provide useful information. Through our process, since it’s anonymous, we hope to increase the quality of information, and since it’s more automated, we hope to increase the quantity and efficiency as well.
Maybe the Thumper rule applies: People who can’t say something nice don’t say nothing at all. Maybe the most meaningful metric would be what percentage of requested respondents actually respond.
That is part of the data analysis we’re able to provide. It’s a feature that definitely has been agreed by our users to be useful.
You talked about giving bland answers in the past as a reference. What’s the blandest response you ever gave?
Answering that question could be quite incriminating for me. Providing any detail to make the story worthwhile would automatically identify this person, and I think that’s just a perfect example of why we are trying to make this anonymous as possible. I think that’s the perfect example of why we need to develop the product the way we are.