The perils of data-driven recruiting
Sorry LinkedIn: you just don’t know me that well.
I’ve used your platform enthusiastically and daily for over eight years, while building out a detailed profile and professional graph.
Despite this, I’ve never once clicked on one of the “Jobs You May Be Interested In.”
Why? The roles are, almost without fail, a comically poor fit for my skills and ambitions.
What’s happening over at LinkedIn? Why is such a core feature so poor?
Despite boasting one of the best data science teams in the world and an unparalleled data set (100s of millions of comprehensive professional profiles), no amount of keyboard matching, scoring, correlation analysis, etc. has been able to solve the problem of consistently matching person to role.
In my case, since I covered the asset management industry as a third-party recruiter, I’m frequently suggested positions within the asset management industry — — most of which I’m completely ill-equipped for.
And I’m not alone in this. Unless you have a clearly-defined role in a well-understood industry, LinkedIn will fail to show you many relevant roles. It’s particularly egregious if you happen to, heaven forbid, be an artist, entrepreneur or similar. It gets worse at senior-levels, where roles get more situationally-dependent and harder to suss out with data.
This speaks to a larger problem in technology today — — the persistent belief that technology solutions are preferable to those driven (at least partly) by humans. We’re told that software is “eating the world” and, as a result, most of the well-funded recruiting platforms have been data-driven solutions. The central premise is usually that improved analysis of structured and unstructured data in resumes or online profiles can effectively inform hiring decisions.
Every year dozens of new startups emerge with some zany variation on the data analysis theme: CalTech-engineered heuristic analysis regimes for candidate discovery! Genetic algorithms to assess the impact of new employees! Salamander-derived neural networks for…. something!
Some notable successes, like Bright, did original and clever stuff, which led to a great outcome — in their case a hefty sale to LinkedIn in 2013.
There’s no question that as more people shift their lives online, the amount of relevant data about their careers is burgeoning and that valuable insights can be gleaned from analysis of this data.
And it’s true: a graduate degree or an impressive work history may be a predictor of success, particularly for roles that require technical skills: A background in finance and a CFA will certainly lend itself more handily to a quantitative investment role than my liberal arts degree.
However, other qualities — often described as “soft skills” — are equally vital in hiring and make all the difference in long-term outcomes. Expertise is not sufficient. Prudent employers know that qualities like raw cognitive ability, the ability to learn, heart, passion, humility and a sense of ownership are the difference between mediocre and star performers.
The Head of HR at a large asset manager recently shared with us that “we’ll take a risk on technical skills and we’ll certainly take a risk on education…. but we’ll never take a risk on cultural fit. It trumps everything.”
Most recruiting leaders are seeking more than just a candidate in a seat. They’re seeking a massive organizational contributor. A loyal colleague. Someone who spends the rest of their career at their organization.
So how do you identify these delicate soft skills? By combining data science with good ol’ fashioned human judgment — likely in the form of referrals.
Thoughtful individuals are wonderful assessors of individual qualities and these are the folks that expensive headhunters rely on every day to get their job done.
At Shortlist, we’re calling these folks Connectors. Our Connectors are senior industry experts who are selected for their judgment and credibility. They understand the industry they work within and, as a result, have a unique insight into the relative merits of different work cultures and whether their friend or former colleague will make a good fit with a certain organization. They offer contextual, position-relevant recommendations that extend way beyond a simple resume. Our clients love it.
Industry Connectors also understand the importance of personal factors and things as seemingly unimportant as timing considerations — we recently had a situation where the successful candidate was someone that our client had interviewed 18 months ago. They loved this candidate, but didn’t re-engage her because their belief was that she would not relocate for the role. However, a Connector recommended her on the platform, explaining that, as a recent mother, she was now open to a role in a non-urban location. She was hired in record-time and the client is thrilled with the result.
We’re not alone. There seems to be an emerging enthusiasm for products that leverage some element of human assessment: witness the rise of Product Hunt, where product recommendations from experts provide exceptional insight into what’s happening at the world’s most interesting technology firms.
Or even the somewhat maligned Apple Music, where artfully curated playlists feel like something tailored by a caring and tasteful friend. Or Apple Music’s 24/7 global radio station, where a live DJ provides a warmer and fundamentally more human listening experience than Pandora’s musical genome.
Recruiting should be the same: unearth outstanding candidates by marrying technology (which can streamline process and identify basic expertise) with human judgement to confirm fit.