It’s easy to poke fun at LinkedIn’s Endorsements feature. Nearly all of us have been endorsed for some marginal skill by some kind soul who knows nothing about that skill or our expertise in it (“Thanks, Aunt Gerta, for endorsing me in…enterprise software?”). Even LinkedIn’s highest paid recruiter platform doesn’t support searching for people based on Endorsements, and recruiters aren’t exactly clamoring for it—everyone knows the data isn’t meaningful.
Maybe LinkedIn hasn’t nailed it yet, but the opportunity is large. Social proof can be extremely powerful. Let’s contrast LinkedIn’s Endorsements experiment with a particularly strong example of social proof already well established in recruiting practices: the employee referral. According to social recruiting research by Jobvite, referrals constitute a whopping 40% of total hires on a small base of 7% of applications, and referral hires stay longer. 82% of employers and recruiters surveyed considered employee referrals the best source of quality candidates.
Clearly there’s something to this social proof thing, but how can we leverage it beyond corporate silos and tap the broader community? As we all know, talent markets are extremely inefficient (in what other markets do brokers regularly pocket 30%?). Even incremental improvements in the intelligence powering talent decisions can have dramatic impact, unlocking trapped human potential.
Whether we’re talking about talent markets or CIA field intelligence, the usefulness of any social proof boils down to the credibility and conviction of the “informant.” Any broad-based attempts to leverage collective social intelligence need to optimize along those vectors, or contend with irrelevance like LinkedIn Endorsements.
Credibility itself is a compound variable that we can unpack into capability to assess, exposure, and intent.
Capability to Assess: Employee referrals are in part so useful because a current employee, as part of the org, is in a privileged position to assess the suitability of someone’s organizational fit. LinkedIn Endorsements, on the other hand, have no tie to assessment capability. Aunt Gerta can endorse me for all the skills in the book (which is not to say she is incapable of any skills assessment, or even exists). Determining the relevant threshold for assessment capability depends on what is being assessed. Technical skill assessment typically requires technical competence (I can’t evaluate anyone’s code), but a person doesn’t need to be “aggressive” to recognize and assess someone as “aggressive”. That said, assessment capability tends to compound (albeit at different rates)—if I asked the most aggressive person I know who the most aggressive person they know is, we’re probably getting to a really aggressive person. Without independently evaluating a person’s assessment credibility, a better social proof platform would do well to offer some transitive weighting to “informants”. The “Python expert” endorsement from a person already endorsed as being a Python expert should have significantly more weight than a Python endorsement from me.
Exposure: Even if someone is well qualified to make an assessment, meaningful social proof relies on sufficient, relevant exposure to the assessed person. Employer referrals don’t tend to intrinsically require a threshold relationship, but we’ll address how intent mitigates that lack of controls. LinkedIn Endorsements require being connected on LinkedIn—a more relevant universe of people to evaluate job-related skills than say Facebook connections, but still quite tenuous. A better social intelligence system would privilege assessments from people who’ve had ample exposure to what they’re assessing.
Intent: Intent is hard to divine, but incentives can help tell the story. As witnesses, whistleblowers trump paid informants every time. I recently spoke with a very successful CEO who retired his latest venture in social recruiting. The idea was to disintermediate recruiting and expand beyond employee referrals by letting the crowd refer people for posted jobs. Anyone referring a hired candidate would get paid for the hire (like a part-time recruiter). Unfortunately, these incentives heavily motivated some of the least credible people to lob in everyone they knew, while the most credible people preferred “doing someone a solid” to overtly capitalizing on an intro. Interestingly, employee referrals also usually offer a cash incentive, but other social incentives are probably stronger. Usually the referrer wants to maintain the relationship with the person being referred, keep face within the organization, and ultimately not invite someone into their community unless they want them there. Thus, employee referrals generally “have integrity” of intent despite the monetary rewards. LinkedIn Endorsements, on the other hand, have no extrinsic reward, but there’s also nothing at stake. The primary motivations appear to be altruism and “tit for tat” (remind me to endorse Aunt Gerta). On par, those aren’t bad incentives—they don’t get too much in the way of integrity, and this sort of gamification has powered a lot of adoption for the Endorsements feature. That said, there’s a much better incentive available. For social proof to really take off in talent markets outside corporate silos, the “informants” need to value the data they’re providing themselves. The best way to ensure integrity in data collection is to connect the user reward to the usefulness of the overall dataset.
Conviction: In social proof, strength of signal is a combination of overall credibility (capability to assess, exposure, and expected intent) and the conviction of the social assessor. The social incentives previously discussed that affect the employee referrer usually cull out the extremely tentative referrer, whereas LinkedIn Endorsers are actively courted to go nuts making endorsements on all their connections. While that gamification may have helped early adoption, it weakened the threshold signal strength for an Endorsement to the point that they were nearly irrelevant. A better social proof platform would take into account different conviction levels. One promising avenue for surfacing real conviction borrowed from the dating world involves enforcing artificial scarcity for better information signaling. Imagine a LinkedIn Endorsements feature where we could only endorse one person as “best” at that skill versus how it is today. How much more useful would that be?
As proponents for anything that systematically helps unlock human potential, we at Talentism would like to see a much better platform for leveraging social proof in talent decisions. This platform would account for credibility and conviction, and it would gather the type of intelligence on people that people want to use themselves. Unlike LinkedIn Endorsements, it wouldn’t focus narrowly on skills. What people are like is inherently more interesting than what people know, and the mirror of others is probably the best way to get that picture. Social proof that someone “doesn’t take no for an answer” or is “relentlessly self-improving” is a lot more interesting than if they “know marketing automation software”. Everyone is interested in learning about themselves and others, not just recruiters. That said, as labor markets continue to shift from the knowledge worker model to the creator model, employers are in increasingly desperate need for this intelligence. Resumes reveal what people have done and what they know, but social proof can paint that richer picture of what people are like. If LinkedIn isn’t going to build it, perhaps we will (VCs can email email@example.com with proposals).