In 2014 three researchers from the faculty of Psychology at the University of Minnesota caused quite a stir with an article in the Harvard Business Review which stated baldly that “In Hiring, Algorithms Beat Instinct”. Their research revealed that “Human beings are very good at specifying what’s needed for a position and eliciting information from candidates, but they’re very bad at weighing the results,” because people are “easily distracted by things that might be only marginally relevant.” The Minnesota researchers argued that machines are better than people when it comes to selecting the right person for the job. And while their final conclusion might have been a little over-the-top, the progress that has been achieved in the use of computer algorithms has nevertheless fundamentally altered the job of the recruiter.

Finding the right person for the job

Finding the right candidates for a job

In the corporate world, the use of complex algorithms and big data analysis has long been the preserve of the Marketing and Finance people. Recently however, Human Resources departments have been making use of this approach, especially for recruitment purposes. Tools now exist to help recruiters at all stages of the process, starting from the search for potential hires – known as ‘candidate sourcing’ in recruiter-speak – to the final selection of the new employee. Some providers, such as the Paris-based startup Yatedo and San Francisco-based Entelo, serve as candidate aggregators. On their website,  when you type the name of a person you are considering for a job, the search engine will come up with all available data. These include not just formal CVs lodged in CV-banks but details gathered from wherever s/he has left a trace on the web – social networks such as LinkedIn and Viadeo, online forums, blogs, etc. 

Tech-based CV verification solutions

However, the majority of the software packages on the market are aimed at the time-consuming area of CV verification. In recent years, the recruitment business has seen the arrival of ‘matching’ sites, based on the principle of social meeting sites. These sites analyse CVs, delete any job-seekers whose profiles are too far removed from the attributes required and prioritise those which best meet recruiters’ stated needs. Explains Céline Memmi, an expert in HR issues at Paris-based Information Systems consultancy Mc2i Group: “Questions for each candidate are listed and weighted according to importance. Each candidate is then given a final score, which enables the recruiter to grade the applications.” The boom in the use of CV verification software is not hard to explain. For mass recruitment campaigns or in cases where companies are swamped by unsolicited applications, it makes sense to automate the initial CV filtering. French startups Qapa and Meteojob have been pioneers in this market. But the heavyweights of the recruitment business, such as Monster, Cadremploi (with its CVaden tool) and the professional social network LinkedIn, have also been investing in this kind of technology, not forgetting the traditional recruitment software programmes such as Taleo, which incorporate this functionality.

‘Predictive recruitment’ claims bias-free results

In a algorithm-based recruitment selection, the failure rate falls to 15%
David Bernard, CEO d'AssessFirst

David Bernard, AssessFirst

Placing the emphasis on the applicant’s personality is also the province of ‘predictive recruitment’ software, which represents the latest trend. These software programmes, which seek to identify applicants whose skills and, more importantly, human qualities, are likely to fit the requirements of their future employer, by studying the proven success factors for a given position, are continually being improved. Recruitment firms such as Talentoday and AssessFirst are popularising this approach. Paris-based AssessFirst started out by seeking to assess applicants’ personalities based on psycho-technical tests. With the advent of big data analysis, the firm has now backed up its expertise in this field with technology capable of establishing a detailed prognosis on the chances that an applicant will prove to be successful in a given job.

Three ten-minute questionnaires suffice to ascertain three aspects: the applicant’s intellectual capacity, his/her reasons for wanting the job, and his/her personality,” explains David Bernard, founder and CEO of AssessFirst. The software then overlays the applicant’s profile on to a company-specific success grid summarising the known success factors for a given job based on historical data drawn up by the company on its current employees: technical tests, personality tests, summaries of assessment interviews, etc. Advocates of this method say that it produces a more reliable result as it is based on a wide range of objective factors aggregated by an algorithm rather than being dependent on an employer’s intuition or mood. “When you observe the results of the recruitment process, 46% of all hirings made in the traditional way, using human subjective judgement, fail within 18 months. When the selection is made by an algorithm, thus ensuring consistent information-processing, the failure rate falls to 15%,” claims David Bernard. The AssessFirst process is designed to obviate any bias based on educational background, social origins or where the person lives, focusing instead on relevant criteria such as character, behaviour and reasons for wanting the job. This is intended to reduce the likelihood of unfair or unconscious discrimination in the recruitment process.

Exploring the chatbot trend

In fact, ‘predictive recruitment’ has been known to produce spectacular results. The AssessFirst CEO tells us about a florist who became a star salesman for the professional tools company Berner. So what are the chances of finding ideal employees from among unconventional candidates? Some scepticism appears to be in order. Underlines Diana Bannholtzer, AssessFirst solution account manager at Fyte: “You still need first of all to find the person’s profile and enter it in the analytical tool. So far, the AssessFirst database is not large enough to search for applicants, to undertake systematic ‘sourcing’ as we call it in our business, in this way within the tool. Moreover, some jobs such as in sales, where it’s paramount to take account of an applicant’s personality and reasons for wanting the job, lend themselves to finding unconventional candidates. But this is not so much the case for fields such as Payments, Legal Affairs and Management Control, where the criteria relating to technical skills are absolutely essential." Meanwhile, the recruitment sector has not escaped the chatbot trend, i.e. the use of software designed to converse with applicants. One example is San Francisco-based company First Job, which has developed Mya, a virtual recruitment assistant.

The Mya chatbot pre-selects applicants for the jobs listed and eliminates those that do not fit the established job profile. The chatbots developed by Do You Dream up and Inbenta can currently answer questions from existing employees regarding the works committee, annual appraisal interviews, paid leave and collective labour agreements. However, they could also be geared up to provide guidance to job applicants who are wading through a company’s administrative procedures.

How is the recruiter’s job likely to change?

Recruitment tools are just filters!

Recruitment tools are just tools

Far from being a threat to recruitment specialists, these complex algorithm-based software programmes can act as valuable assistants when it comes to the first filtering of job applicants. Head-hunters, whose job used to be painstaking and very time-intensive ten years ago, are now saving huge amounts of time thanks to automation. However, despite the innovations taking place in this sector, no-one imagines that the selection of a new recruit could today be handed over entirely to a machine, even to an advanced solution such as predictive recruitment software. “Predictive recruitment merely provides an additional indication to help choose the right person,” stresses Fyte’s Henri de La Roque. On the positive side, the range of software available has certainly broadened the recruiter’s capabilities. Now freed from the more onerous, time-consuming tasks, s/he is able to focus on the human aspect and emphasize his/her advisory role. “Knowledge of the ecosystem, negotiating, talking through the job with the applicant, telling him/her about the history of the company… a recruitment consultant plays a vital role in creating the employer-applicant relationship,” points out de La Roque. And it will not be easy to replace human beings in this essential role by a robot, however sophisticated it might be.

By Olivier Discazeaux