Current Issue
This Month's Print Issue

Follow Fast Company

We’ll come to you.

4 minute read

LinkedIn's Algorithm Taps Talent Graph, But Still Needs Human Touch

Imagine if your prospects for beating the 9.1% unemployment rate depended not on a meticulously crafted cover letter and résumé, but on a complicated algorithm that helped companies determine the best matches for open jobs. Such a Brave New World-like future is rapidly becoming a reality—but don't fear just yet. Your career will still very much rely on strong credentials, networking, and a good pinch of serendipity. (For now, at least.)

Over the last year, LinkedIn has rolled out a set of new premium tools to its 100 million users. And they've worked. On the consumer end, what would normally cost businesses and HR departments time and money now just takes a few clicks. On LinkedIn's end, the network is gathering so much data, says Adam Nash, VP of product and user experience, that it's "starting to really get an understanding of who the best are in their fields, and more important, who are the best fits for your team." Internally, the company refers to LinkedIn's search algorithm as the "Pandora for people," a system that combs through the network's "talent graph" for ideal job candidates. (Not another Pandora analogy!) When it works like it should, all employers have to focus on is the final interview. 

"We're starting to see recruiters do queries where they literally put, 'I want someone who has worked at one of these twenty companies and a startup, and gone to one of these twenty schools,'" Nash says. "You're never going to get that from a résumé."

Two tools in particular are helping businesses find talent: Skills and Similar Profiles. Launched in late June, the latter feature enables employers to discover new talent based on profiles of top workers or ideal candidate profiles. "You just tell LinkedIn, 'Look, I have five great engineers, and I want more like them,'" Nash says. "And we just find people like that for you." Skills is an even more in-depth feature—it lets companies narrow down candidates based on talent and influence. In other words, it's a Klout score for the job world.

"With Skills, you're never going to see a score," Nash clarifies. "What you see on Skills is something that we internally call the 'talent graph,' an understanding of what the skills are, who has them, and who influences each other. In some ways, I don't want to know whether you influence a lot of people. If you're a great [programming framework] Ruby On Rails engineer, I want to know whether you influence other great Ruby On Rails engineers. I want to know whether you influence them about Ruby On Rails. I don't want to know whether you influence them about politics."

He continues, "There's a lot of talk on the web of the interest graph: fans of this and that. But what's really interesting about Skills and the talent graph is once you really understand what these skills are and how they're related to each other—how they're related to people and their relationships—you start to get much deeper than a Klout score." Nash believes this data will make for "next-generation" applications on LinkedIn.

But as bountiful as LinkedIn's data is, it's not the be-all end-all. "There's always going to be a lot of value for certain types of positions in what I'll call human intelligence," Nash says. "Because let's be honest: Even if I give you the perfect 10 candidates, I don't care how great that automated system is. There's still that final eye for all those subtle details—there's really an emotional decision that's made." He recalls that Reid Hoffman struck a similar tone when he first met the LinkedIn founder. "Joining a company isn't a one-time transaction: It's not like buying a car, where you negotiate the deal and then you're done," Nash recalls Hoffman saying. "It's a long-term relationship. Your happiness at that company isn't about what you agree to that first day. It's about that next month, that next year."

Indeed, that human understanding played an important role during Hoffman and Nash's first meeting. Nash, who had been at eBay for the previous four years, was considering leaving for a new position at another company. A friend suggested he connect with Hoffman for advice.

"Reid and I ran in the same circles for more than a decade, but never met," Nash recalls. "We met for breakfast at a local spot called Hobee's. It was supposed to be an hour breakfast. We ended up talking for four hours." The conversation flowed from discussions of product management to the web 2.0 world to how best to build organizations of great people. After hours, the pair finally left.

"We were walking to the car—I remember it clear as day," Nash recalls. "He said, 'Well, you know, I'm involved in a lot of companies, but if you're really interested in this stuff, we are hiring a senior product person at LinkedIn."

By Sunday, Nash had an official offer from LinkedIn and started at the company only weeks later—a hiring that likely would never have happened without a human connection. Not that LinkedIn didn't play a role in fostering their relationship.

"But here's the funny thing," Nash says. "Once I got to LinkedIn, one of the things I learned we track internally is who invited who [to the network]. If you joined LinkedIn, somewhere [in the graph] it says who the one person is who got you to join LinkedIn. We track the generations over time. There's that first generation which were all the invites sent by the founders—it literally goes back to that. I found out that I'm second generation on LinkedIn."

But more significantly, Nash discovered his connection to Hoffman. "Reid invited my friend John Lilly, who is now a partner at Greylock and who was the CEO of Mozilla. John Lilly invited me to LinkedIn, back in 2003," Nash says. "And who introduced me to Reid for that breakfast years later? John Lilly."

[Image: Flickr user RedCraig]

loading