What if it were possible to uncover just the right person for a job, one that will be a high performer, who will blend into the team as if she had always been a part of it and who will not jump overboard just two years later? Even better: what if their profile would appear to us without having to dig through hundreds of CVs? Welcome to the world of “predictive recruitment”.
“Predictive recruitment is entirely possible today,” says Andrée Laforge, product manager at Syntell, a firm that specializes in business solutions for human resources. “However, in practice few companies are currently able to do it.”
“We have the technology, but we do not have either the quality nor quantity of data to take advantage of it,” she adds. In Quebec in particular, there are very few companies who possess a sufficient amount of data for the algorithms to give reliable results.
Hunting for the perfect candidates
Syntell recently assisted an organization that wanted to slow down its turnover rate of its customer service agents. Many were leaving after a few months while some remained for several years – what was special about them?
Two other criteria were added to the research to identify the ideal future employees of the organization: not only should they have been on the job for at least two years, but they had to perform well in sales and collect positive assessments by customers surveyed.
“Demographic data, places of residence, academic record, personal interests… we collected everything we had in the way of data on these people to identify common characteristics,” says Andrée Laforge.
What about discrimination?
Think about it, the algorithm chooses the profile of the ideal candidate: men from 25 to 35 years old, of Latin origin, with a high school diploma. Should the organization automatically exclude all the women who apply? “Of course not,” the data specialist replies. “That would be pure discrimination. Be careful. The goal here is to direct the posting of positions to be filled to sites or publications that address this target group. There is never any question of excluding applications.”
Is the line too thin? Andrée Laforge doesn’t think so. “Robots do not discriminate! They are much more objective than humans. In human resources, we work a lot with relationships, the famous “pif”. However, the data exclude the subjective. Perhaps we can finally convince this manager who studied at McGill that a candidate from the Université de Montréal deserves his attention?,” the specialist gives as an example.
In the grey area
Although it is difficult to collect a lot of data internally, through the LinkedIns and Facebooks of this world, it is not volume that is lacking.
“So we talk about scraping, another means used by predictive recruitment, which requires much more care,” explains Andrée Laforge. “In theory, we do not have the right to build a database from information provided on these media.”
In practice, every headhunter in his time has done it by hand at least. For example, he will go through LinkedIn profiles extracting “data analyst” from the enormous directory, then look at Facebook to build an idea of the individual’s personality, before contacting them or refraining from doing so. “If we exclude a person based on one of their personal pictures, for example, this is discrimination,” warns Andrée Laforge.
In fact, no one will know. If this criteria is instead expressed in an algorithm that a computer uses, a trace is left behind. So will automating recruitment reduce discrimination on hiring?