The fall of human recruiters.

Ibanga Umanah
Brave
Published in
6 min readJan 17, 2019

A short history of tech’s unbundling of the hiring process.

pixabay.com

How did recruiting start?

Soldiers in the European empires of the B.C. era got a bonus for bringing their friends — what we now call referrals. Forward many years later, and recruitment advertising appeared as an alternative to local networking. Recruitment advertising helped to cover the gaps left by sending people off to war. For a long time that’s how it went: networking and advertising.

Recruiting is an old industry.

Like most old industries, it largely operates on tradition. I hire people the way the people before me did, which is the way the people before them did, and so on. Most job boards look like newspaper classified advertisements, not out of any practical insight, but out of dogmatic tradition. And like most dogmas, the relevance of its ideas expires as the world changes — which is every single day.

Referrals require a network as diverse as a company’s needs, which is highly unlikely, given the pace of change. Advertising, on the other hand, depends on people self-selecting for a company’s preference correctly. This doesn’t usually happen as high unemployment rates lead people to apply for anything remotely tangential. This disconnect between hiring dogma and the real world results in recruitment errors that reach 50 % across the industry.

The nearly $60 Billion recruitment industry is made up mostly of dogma players, with only a few small experiments. It’s also fragmented. The largest player — LinkedIn — makes $2B from its recruiting efforts. The next big 5 recruiters make $500M — $1B each. This translates to less than 10 % market share distributed among the top ten players. Over two-thirds of the market is dominated by thousands of $500K, 10-person consulting services.

Both big and small consulting services usually offer to outsource the entire process of recruitment. However, their scale is tied to their geographic footprint, with most recruiters specializing in a particular city. To get bigger, they have to hire another recruiter in a different city.

With the advent of the internet, technology companies started to unbundle these human recruiters into specialities. Generally, we’ve found three:

  1. Sourcing: Finding people to hire.
  2. Applicant Tracking: Managing the hiring process.
  3. Recommendation: Deciding who to hire.

Finding People

Tech started with finding people. In the early 1990’s, job boards replaced newspaper classifieds and recruiter rolodexes. By the early 2000’s, social networks became a competitive alternative. Very little has changed since then. New entrants have focused on increasingly niche communities like HackerRank and StackExchange for developers as well as Dribble and Behance for creatives.

Competing in this domain requires developing a relevant niche, building a brand community, and creating a network effect. Because these are often winner-takes-all plays, the market is not only saturated but ultimately very expensive to win. A recruiter’s personal network isn’t as valuable anymore.

Manage the Process

Shortly after the online job board revolution, Customer Relationship Management systems started improving hiring logistics — storing contacts, pulling up resumes, scheduling interviews — a set of tasks known in the industry as applicant tracking.

These followed the natural progression of software becoming a service (SaaS), with more recent innovations focusing on user-experience and performance tracking. So it stands to reason that the best people to build a competitive company in this area aren’t scientists, but designers and software engineers who understand process and data. For example, the two newest and most popular Applicant Tracking System (ATS) platforms, Lever and Greenhouse were founded by a data visualization designer and a software design consultant. That said, some of the most well-known tools come from SAP, IBM, and Salesforce, who focus on company operations rather than exclusively recruitment. It’s not too difficult for these behemoths to build in recruiter best practices into their process, making human consultants obsolete.

pixabay.com

Deciding Who to Hire

However, the only way to decide who to hire has been through an in-person interview; the last stronghold of the human recruiter. So arguably, given the breakdown of the $60 Billion market we discussed earlier, deciding who to hire represents about two-thirds of total recruitment spending.

Naturally, tech wants to do this too. With the increased popularity of relational or network-based search (such as Google) and online tracking (such as the Cookie), people have been working to build better targeted recommendations.

However, if you’re looking for talent, you can’t just google it. At least, not yet.

Keyword recommendations are exceptionally poor when the variety of jobs and people are nearly infinite. Additionally, a study by Schmidt et al. [1] on employee performance found that recruiter’s go-to information — experience — predicts about 5% percent of employee performance. The heterogeneous and continuously evolving nature of people at work combined with the shockingly lax nature of social science has made automated recommendations about as good as a coin flip — if not worse.

That is until recently.

A mix of neuroscientists, psychologists, and Artificial Intelligence (AI) experts have entered recruiting to help facilitate filtering and choosing. For example, a machine learning algorithm can identify skills and aptitudes a person may have, even when they don’t explicitly appear on a candidate’s resume [2]. At Google, scientists are hard at work trying to understand their own biases during in selection [3].

Despite progress in reducing both noise and bias during talent selection, no one has been able to achieve an accurate recruitment recommender system. People are complex and so are companies. Not only are they full of nuance, but they also change over time. So whatever model you start with will also need to change.

For now, human recruiters still have a job.

We’re not tech utopians or particularly attached to AI taking over everything. It’s simply that by and large, the way recruiters work, even after reducing bias, is still fairly hit or miss. As the psychologists Kahneman and Tversky have shown, human predictions just don’t work that well [4].

Without good prediction, companies and the smart people in them make less impact than they otherwise could. We need your help to change this.

Get in touch as this isn’t a problem we’ll solve alone.

Authors

Ibanga Umanah is a Cofounder and the Head of Strategy for Brave Venture Labs. Brave is a people science company uncovering the drivers of performance for better recruiting and talent management.

Amina Islam has a Ph.D. in engineering and is currently putting her skills and academic background into doing evidence-based research on the impact of informal learning programs.

Get in touch to hire, get hired, or join our team: brave.careers

All images courtesy of pixabay.com

References

[1] F. Schmidt, I. Oh and J. Shaffer, “The Validity and Utility of Selection Methods in Personnel Psychology: Practical and Theoretical Implications of 100 Years of Research Findings”, Researchgate, 2016. [Online]. Available: https://www.researchgate.net/publication/309203898_The_Validity_and_Utility_of_Selection_Methods_in_Personnel_Psychology_Practical_and_Theoretical_Implications_of_100_Years_of_Research_Findings. [Accessed: 21- Dec- 2018].

[2] N. Scheiber, “A.I. as Talent Scout: Unorthodox Hires, and Maybe Lower Pay”, Nytimes.com, 2018. [Online]. Available: https://www.nytimes.com/2018/12/06/business/economy/artificial-intelligence-hiring.html. [Accessed: 21- Dec- 2018].

[3] “re:Work — How Google uses data, structure, and science to hire”, Rework.withgoogle.com, 2018. [Online]. Available: https://rework.withgoogle.com/blog/Google-uses-data-structure-and-science-to-hire/. [Accessed: 21- Dec- 2018].

[4] D. Kahneman and A. Tversky, “On the Psychology of Prediction”, Psychological Review, vol. 80, no. 4, pp. 237–251, 1973. [Accessed 21 December 2018].

--

--

Ibanga Umanah
Brave
Editor for

Building Brave Venture Labs // here to learn