It seems inevitable; when applying for a job became as easy as a mouse click, companies were inundated with a tidal wave of applications. It was a happy confluence of events, labor markets were expanding, and businesses around the world were increasingly taking their presence online. HR departments could finally streamline and automate some of their most laborious activities.
The original promise for applicants was an easy sell– no longer was it necessary to trudge through job openings in the paper, call receptionists, or in most cases, walk into an institution and hand someone your resume. Now, all you needed was an internet connection if you wanted to land your dream job. Similarly, recruiters rejoiced in the fact that they had reduced the time spent head-hunting or screening fruitless phone calls, and could re-focus this newfound time hand-selecting the top echelon of candidates that had sought them out.
Unfortunately, pushing the recruiting process online meant more competition for job seekers and subsequently, an infinitely larger number of applications to sift through for recruiters. That time that was supposed to be spent engaging with creme de la creme was funneled into reviewing thousands of applications via an online portal. It was a sure-fire way to ensure recruiter burnout, and inevitably, the ever-present danger of missing out on amazing candidates.
It was the problem that AI was supposed to fix. Apply intelligent systems to pluck out the best of the best in an ever-expanding database of applicants and remove the potential of human error or bias. Cut to today, hundreds of tools promise fancy algorithms that sort through the hoard of applicants to find the "VIP candidates," utilizing resume screening, video assessments, reference checks, and numerous other predictive tools. Supposedly, accurately mimicking the very personal activity of finding, meeting, talking, and sourcing the right people.
However, as we've seen repeatedly, coupling automation and scaling in hiring does not work.
Automation doesn't scale hiring.
For us, it's a tale as old as time– automating the recruiting process is a quixotic attempt. As we've highlighted here before when you pair automation and scaling to vertical hiring marketplaces (platforms that cater to one specific demographic), issues arise. For those of you who have only recently joined us, here is a quick refresher:
In the last five years, we've seen all sorts of these platforms, from Triblebyte to Markterhire to Hired receive venture funding, with the expectation of scaling their products. When you start with 500 front-end engineers, the process is quite smooth. You can vet each one with some sort of proficiency quiz, log personal information more authentically through 1:1 contact. Effectively, you can ensure quality on the talent side.
Your business, at this stage, is people– understanding them, figuring out what kind of work environment they'd thrive in, how good they are at their craft, it's a high-touch business.
But once you start scaling, your business becomes lots of things. You need to focus on optimizing – building out your website funnels, user acquisition, retention, and all of the sudden, there are misalignments in cost and value structures. Instead of doing what you do best (vetting and placing), you're focused on making sure you're reaching the right demographics to scale quickly.
Now you've spent a boatload of money hiring growth marketers and focusing on optimizing your landing page and its achieved what you wanted— 500,000 front-end engineers. The problem is you've trapped yourself into the same negative network effects you find on LinkedIn. Eventually, participation in the marketplace means very little, and recruiters and companies are essentially forced to rely on the same signals they'd rely on anywhere else— third-party institutional stamps of approval. You've lost sight of the high-touch vetting process that made your platform so valuable in the first place.
AI is not the answer to recruitment platforms' issues with scaling (even though it may look like it). No matter what, to sift through the sea of applications and get the best applicants, you have to establish some type of credentialing system. When you try to effectively de-risk with AI recruitment tools (remember recruiters always want to reduce risk, the cost of a bad hire can be cataclysmic) what happens is a closed-loop system that reinforces traditional credentialing systems. This is one way it can look:
Ads created by algorithms encourage certain people to apply for a job –> The resumés go through an automated culling, where the system looks for certain institutional stamps of approval or the right keywords on a resumé –> A lucky few are hired and then subjected to an automatic job evaluation –> These results are looped back in to establish criteria for future job advertisements
When AI is applied to recruiting, it presents a host of additional issues too, whether it's gender or racial bias, outdated institutional stamps of approval, or attempts to game the system.
Sure, AI might remove busy work and some human bias, but machines introduce a more severe and unforgiving form of bias. These systems also totally ignore the fact, as we've written, that professionalism is leaking out to every part of our lives, well beyond what a single page resumé can hold.
Optimizing your product funnel through a very limited scope might have been ideal historically, but the reality is, job search and recruitment has accelerated in informal pipelines. Water cooler conversations and networking events are now Twitter DM's, job interviews are shout-outs on Discord or a DM in your slack writing community.
Social communication and professionalism are unifying in a way that we've never seen before; the signals are become increasingly informal, relational, and distributed.
Hiring is human.
AI can only go so far in sourcing the perfect candidate for your role, the act of actually deciding on a candidate relies on uniquely human social activities.
Let's ask a question:
You, as a recruiter, are presented with two candidates; Candidate A and B. Let's say candidate A has a glowing resumé, graduated Ivy League, and worked for 3 out of the big 5 (FAAAM); on paper, they are a superstar. Candidate B by comparison, has a decent resumé, didn't graduate Ivy League or work at any of the big 5, but has relevant experience at a few startups that would make them a good fit for the role. They've both interviewed with you in person, and they were equally impressive. You've checked their references and no red flags on either account.
Now, aside from the fact that an AI system might have overlooked Candidate B entirely, you have to make a decision. But would you stop there? No. You do your due diligence, you dive into your network to see if you and each candidate share any similar colleagues, and luckily you do– better yet, a current colleague of yours has worked with both of them. You reach out and ask for their take on both candidates. He shares with you that while Candidate A has a great resumé, they are known for creating toxic work environments everywhere they go. Candidate B, on the other hand, while his resumé is not as stacked with brand names, is a pleasure to work with and repeatedly helped foster positive and productive work environments.
Hiring is a multi-factor decision, but would this last stop in your process not influence your decision in a far greater way than nearly any other data point you've collected thus far?
Additionally, wouldn't it be entirely more powerful to start your candidate search with this information at the outset? Most recruiters would much rather receive candidates from referrals anyway– they are quicker, cheaper, and tend to stay at companies longer. Additionally, referrers tend to be more engaged when they make a successful referral– quite simply, it makes them feel good.
Referrals will forever be the most cherished point of evaluation in the hiring hierarchy because they carry the most weight amongst all candidate signals. The referrer assumes risk of social capital when they make a referral. By referring someone (a real referral, not one on LinkedIn), you are putting your own credibility on the line, sticking your neck out on behalf of someone else. A lousy referral can damage your professional and personal reputation in the same way, that a good referral can boost your professional and personal standing. Unlike AI, which assumes no risk and places all the risk squarely on a business, it doesn't cost anything for the AI to make a bad decision.
No matter how much automating and optimizing you do around credentialing, referrals, whether from a colleague or someone within your network, will always be the clearest and most valuable signal to a decision-maker. Referrals, not candidate volume, are what so many recruiting tools have missed; they are scaling the wrong data points.
We've seen how powerful referrals and network chains are in other social products. Aren't you more likely to accept an Instagram friend request when you see 20+ of your friends also follow that person? When you match with someone on Hinge, doesn't it put you at ease to know that you share mutual friends with that person? Someone you could reach out to who could provide more info on the type of person they are?
Common points of mutual connection make for stronger introductions and engagement. As we see professionalism and social communication collide– network chains and referrals are becoming exponentially more valuable for recruiters and companies. It's the same reason small-time recruiting agencies are still around. They've centered their whole existence around the axiom that hiring is human. The ultimate goal in recruiting is to find the people that can say, "I have people, and I know what they'd be good for."
We believe in the power of networks. When a company is looking to hire a position of significance, they don't want 10,000 people through their door via an AI algorithm. They want qualified candidates, and in the best case, qualified candidates are referred by someone's network.
Our thesis has always been that hiring is a human activity. We believe the uniquely human aspect of hiring, referrals, have thus far been limited to immediate networks. Like a singular node, recruiters have been siloed to utilizing their own networks to source referrals, but what if they could see every single node on the network itself? Better yet, what if by virtue of being online and consuming content you could have access to every network there ever was?
With a product that digitizes the process of referrals, you effectively cut out the credentialing-AI gatekeepers. Give this power instead to creators and natural audience builders, and you'll build a totally transparent social web, one where trust goes further than one degree, and one that's not longer shrouded under mysterious algorithms.
Hiring is human, and it's going to stay that way.