the algorithm determines if you get the job


Image EPA

Team leader Charlotte welcomes applicants in the hall of the Arvato call center in Venlo. After a short tour, she puts the upcoming phone employees to work with a smile. They write emails with the greatest concentration, helping fictitious customers who call dissatisfied and solving geometric puzzles. ‘That intelligence test was pretty tough, right?’ Charlotte asks afterwards in the cafeteria where two boys are playing Fifa on Xbox. ‘All tests have been performed. Well done. Thank you for joining us. ‘

The application takes about an hour and a half. But the trip to Venlo is spared for the job seekers. They follow the trip with Charlotte from home, behind the computer. They also answer the questions online and they complete the tests that Charlotte takes on the computer. The application is part of the online game Talentpitch. The main prize: a job at the Arvato call center.

Computers are increasingly taking over the human resources departments. Instead of humans, it is algorithms, mathematical formulas embedded in software that identify the perfect candidate from a range of applicants. By analyzing the applicant’s language, using games or psychological tests, a computer collects data about the candidate. After a calculation, a ranking is rolled out with the most suitable candidates and their scores.

“It’s then up to the managers to decide if they really want to hire the candidate chosen by the computer,” says Barend Raaff, CEO of the fast-growing software company Harver, which has designed the application algorithm. Raaff: ‘Arvato was our first customer. The company came to us four years ago with the question: What are the predictive characteristics of a good call center agent? At that time, 70 percent of Arvato’s staff left within the year. That turnover was going down. ‘ Harver then developed the Talentpitch program with Arvato.

Harver now has offices in Amsterdam, London, New York and Tel Aviv and a customer base in 22 countries. “The call centers of big companies like Netflix and Booking.com are now working with Talentpitch,” says Raaff. ‘Retail chains like Rituals and MS Mode use it to hire store staff and catering companies to select employees in the service or behind the reception.’

language use

Multinational companies such as L’Oréal and PwC are also in line with the Dutch-Chinese company Seedlink. Seedlink also has software that can predict which applicants are best suited for a vacancy. ‘We analyze the applicants’ language use based on their answers to three open-ended questions,’ says co-founder Rina Joosten-Rabou from Shanghai. The machine brain then predicts whether graduates score well on personality traits such as leadership, customer friendliness and creativity.

Both Raaff and Joosten-Rabou say their software makes it unnecessary to read resumes and conduct interviews. ‘CVs can not predict whether someone is customer-friendly and stress-resistant, and in a job interview you can not say whether someone who wants to become a call center employee has good ear-hand coordination,’ says Raaff. ‘In addition, the HR managers who evaluate the CVs and hold the interviews take into account their own prejudices,’ adds Joosten-Rabou.

Seedlink and Harver’s self-learning algorithms should reduce human influence and their prejudices. Because the employees who have worked in the company for a long time also fill in the open questions and play the games, the computer can compare the applicants’ results with the employees who perform best. The top applicant will then float to the top.

Small business

Now the technology is primarily available to slightly larger companies. ‘A company must have at least fifty employees for a tailor-made analysis,’ says Joosten-Rabou from Seedlink. The company builds standard models for an entire industry. Joosten-Rabou: ‘This allows the algorithm to predict the success of a small business salesperson by comparing his scores with those of salespeople at other companies in the industry.’ This should ensure that the technology not only remains a toy for multinational companies but can also be used by small businesses. ‘The hairdresser on the corner will also use it in 2019’, says Joosten-Rabou.

Tilburg professor of staff and organization Jaap Paauwe doubts: ‘Only if the hairdresser is very computer-oriented and happens to have the right contacts. In most hair salons, only the best interns will move on to a job. It’s still the best way to see if anyone fits. ‘

Yet Paauwe also believes that digital selection methods for large companies contribute to a better match. ‘While there is as yet no evidence that they really work better than a cover letter or an interview, they will certainly be used more and more with a larger number of applicants. But man will not disappear completely. The algorithms are especially useful as a first round. The companies will first have a conversation with the best candidates. ‘

This is also evident at the Arvato call center. After the applicants have passed the Talent Pitch, those who have been selected as the best of the education will be called up. If it then clicks in a short conversation, they can call themselves Arvato’s phone employee.