“The computer says you’ll probably work well with the humans at this firm.”

If you think that sounds like a joke, think again. Large businesses have been trying to partially automate the hiring process for years, with CV-scanning systems and computerised testing that can sometimes filter out candidates with minimal human oversight.

But now organisations are increasingly asking algorithms to assess whether a person will be a good “cultural fit” at the firm. That doesn’t – necessarily – mean people will get on socially with their colleagues, it means their way of working will suit the organisation at large. How can a computer predict that?

At London-based start-up ThriveMap, a consultancy offering pre-hire assessments of prospective candidates, the answer is to use interactive questionnaires that “simulate” a day in the life of a new employee.

ThriveMap’s clients are generally large firms, where job opportunities can attract hundreds or thousands of applications. “You can’t offer a trial shift to every candidate that applies,” says chief executive and co-founder Christopher Platts.

However, it is possible to ask candidates a series of questions digitally, via video clips or written text. At ThriveMap, these queries and scenarios are specially designed for each client and are intended to probe how the job candidate would approach their work in a specific role.

For instance, the successful applicant might have to get up early to start a shift on time or deal with a higher workload when the weather is bad. How do they feel about that and how would they go about it? The survey might also ask detailed questions about how the candidate would attempt to solve a problem should something unexpected happen.

Such questions are about competency but also cover professional attitude. They could simply be asked during a face-to-face interview, but Platts’ big idea is to bring them up within simulations so that candidates can be screened at an early stage of the job application process.

Employers might as well start selecting for cultural fit as soon as they can, argues Platts. And he says the results speak for themselves: “Candidates that score ‘excellent’ or ‘good’ on our assessment are three times less likely to leave in the first 90 days.”

A scoring system is agreed with each client in advance, so that the algorithm can determine how well a candidate has answered a given question. The tool might also deduct a few points should the candidate respond very slowly to questions, for example. But Platts says there is a “full audit trail” so that the decisions made by the system can be traced afterwards if necessary.

Social media search

Part of the challenge in finding the right person for a job is getting the ideal candidates to apply in the first place. Data analytics is helping source prospective employees based on cultural fit in this way, too, says Marilyn Tyfting, chief corporate officer at Telus International.

The company (a subsidiary of Canadian telecommunications firm Telus) hired more than 25,000 people in 2019, with 30% sourced from social media, including Facebook and Instagram.

Telus International’s employees provide phone support to the customers of third-party clients. For example, a company that makes video games might pay for Telus’s staff to help gamers who are having trouble completing a new game.

For that kind of role, Telus wants employees who have gaming experience themselves. The company uses an automated system to flag up social media profiles that indicate a person has some connection to gaming. That could be based on the accounts they follow or the links they post. If the individual’s social media account is open, they can be contacted directly. If it’s private, Telus might send them a private message to ask if they’d like to discuss work opportunities.

If someone decides to go ahead and apply for a job, they’ll be asked to complete a couple of proficiency tests. All of this is organised automatically. Should they score well, the system might pass their details on to a human recruiter who can then offer them an interview.

Sometimes, prospective candidates actively seek out jobs at the company via social media themselves, says Tyfting, so it works both ways. “People find out who we are and choose to comment and interact with us,” she explains. “That on its own indicates a certain amount of cultural affinity to who and what we are.”

Scouring social media for candidates who have a mindset that will suit a particular employer has become popular with tech companies. Some, for instance, use algorithms to hunt down profiles on the code-sharing website GitHub and then offer promising individuals the chance to apply for a particular role.

Again, this isn’t just about finding people with the appropriate qualifications or skills – it’s about the way they approach their work, too. Could that person be described as a “thought leader”, or someone who is clearly interested in keeping up to date with new technologies? An automated system could work that out by assessing how influential an account is among its followers – or how regularly it mentions a diverse range of programming topics, for example.

Tracking a team-player

Using data to find candidates who will be a good cultural fit has its advantages but it isn’t without pitfalls, says Tom Calvard at the University of Edinburgh Business School.

If companies set criteria that are too rigid, they could end up hiring the same kind of person again and again, turning the working environment into a monoculture.

“Over time, you hire in your own image or an idealised image and everybody becomes too similar,” says Calvard. Applicants might even get a sense that this is the case before or during the hiring process, which could leave them doubting that the job is right for them because the “ideal candidate” seems so specific and homogenous.

If organisations begin emphasising the importance of “cultural fit”, there is indeed a danger that job candidates and existing employees will end up trying to conform at all costs, says Sameer Srivastava at the University of California, Berkeley’s Haas School of Business.

He and colleagues have recently studied how to detect signs of cultural affinity in the text that employees write to one another in email messages and Slack chats. The researchers have also analysed how employees manage their digital calendars and respond to event invitations.

“When you’re writing just to colleagues, what happens is that elements of your cognition, personality style, subconsciously come out in the ways that you write and communicate,” says Srivastava.

By using a text analysis method called Linguistic Inquiry and Word Count (LIWC), Srivastava and his colleagues say they are able to tell how well-fitted an employee is based on how they use personal pronouns in email messages, for instance. Do they signal team awareness by referring to work that “we” are doing – or do they rely on “I” and “me” a lot?

By assessing language in this way, Srivastava and other proponents of LIWC analysis say they can tell whether someone fits in naturally within a group – but also whether they adapt well over time when the group’s dynamics change. This ability to be plastic, to adapt, is often what organisations should really be looking for, says Srivastava.

His colleague Matthew Corritore at McGill University agrees. In a recent paper, he and co-authors described how employers might be at their most competitive when they hire employees who can both fit in well, which is good for efficiency, but who also have the capacity to challenge norms in the working environment and adjust to changed circumstances when necessary.

“You still need some shared common ground but to be innovative you want to make sure that you yourself, the average employee, is bringing novel ideas about how to do your work,” he explains.

Corritore and his colleagues say that this within-person diversity is also something that can be picked up in how people use language day-to-day.

It raises the possibility that one’s cultural alignment and adaptability could be assessed not just at the point of applying for a job – but also continually, via software installed on company computers and smartphones. While that could help sharpen corporations’ competitiveness, employees might also disagree with the algorithmic judgements being made about them, says Calvard.

“In the next 10 years, we may start to see employment tribunals that revolve more around tracking and algorithms – and how that data’s been used,” he warns.

 

Compiled by Olalekan Adeleye

BBC