Why you're terrible at recruitment (and can AI help?)

The short version is you're full of biases and your hiring processes are badly designed. Fortunately, there's something you can do about it.

by Emily Burt
Last Updated: 15 Jan 2020
Also in:

Here are a few interesting facts to consider before you start looking for that high-flying new recruit. If a candidate’s name lies towards the end of the alphabet, it might not matter how sparkling their CV is – they’re likely to be at a disadvantage. If you’re hungry when you’re considering applications, your judgement may well be impaired. And no matter how open-minded you believe yourself to be, you have almost certainly judged an interviewee on a whole range of qualities and demographic certainties within seconds of meeting them.

It’s little wonder that we so often end up hiring the wrong people. According to a 2018 global survey from recruiter Robert Half, more than eight in 10 HR decision-makers admitted they had made bad recruitment decisions, and 39 per cent of them realised it within two weeks of an individual starting work. Numerous studies have attempted to quantify how expensive such errors are and have put them at anything between £30,000 and £130,000 depending on the seniority of the individual involved. 

Part of this can be attributed to bad luck or, in the UK particularly, the inevitable scarcity of talent in certain disciplines and sectors. But the bigger picture is that despite a recruitment industry that has developed into a multinational behemoth, the collective wisdom of generations of behavioural scientists and organisational psychologists and a range of highly impressive technology, most of our hiring decisions come down to whims and hunches. 

It barely needs to be said that this is almost always the wrong way to recruit. The question is whether we are ready to admit it.  

Tomas Chamorro-Premuzic, psychologist and author of The Talent Delusion, says both individuals and the broader industry are culpable. "Decision-makers and HR professionals really overrate their intuition, and when it comes to people decisions, the majority of people believe a short-term interaction, whether it’s a video interview or even a CV, can predict whether someone will be a good fit in the organisation," he says. 

"But the science is complex and academics historically have not bothered to make the science around hiring digestible or accessible to the broader workforce. As a result, you get the spreading of recruitment ideas that often don’t have a factual basis but are relatively easy to follow and replicate."

The idea that bias underpins the majority of our hiring problems has at least reached the mainstream, but few of us understand the scale of the issue. In a 20-minute meeting, for example, an individual transmits up to 700 non-verbal cues, which means the things we think are important often form only a small part of the picture.

"As soon as you meet a person, you peg them, unconsciously, on up to 150 different stereotypes," says Tara Swart, neuroscientist, leadership coach and author of The Source: Open Your Mind, Change Your Life. "The obvious ones are gender, age and race, but it will also encompass social class, wealth, political affiliation, accent, educational levels and physical attractiveness. If you’re a well-regulated person emotionally, you should be open to reversing those first impressions, but they are a staple of the way we interact." 

Bias is an uncomfortable topic, but it is natural and inevitable, rooted in instinctive ‘System 1’ thinking without which we would be incapable of functioning. To prevent itself from collapsing like a black hole under the pressure of billions of stimuli every day, the brain is forced to undergo a rigorous and lightning-fast filtering process. 

The amygdala, a collection of neurons, frantically processes decisions on which cues and feelings to focus on. Thanks to the sheer volume of information, it doesn’t prioritise the most comprehensive interpretation of this data, but the fastest and most efficient. The hippocampus makes sense of it by pairing new information with subjective memories, steering us towards one option over another, forming opinions that feel objective but are nothing of the sort. 

This process has a tendency to manifest in unacceptable ways. A 2016 study on racial biases in the labour market found that when black and Asian candidates removed indications of their race from their CVs, they were invited for a greater number of job interviews, while a 2017 BBC investigation found that identical applications filed under the name Adam received three times the number of interviews as Mohammed. 

Bias can also be a lottery – merely the time of day an application lands on the desk of a recruiter could be enough to skew the candidate’s odds, according to one study from Columbia University. 

Irritatingly, unconscious bias is not something you can dismiss altogether – research has shown first impressions can predict real-world outcomes. In one study, subjects were shown a sheaf of photos of company CEOs, and the amount of response activity in their amygdala was found to accurately correspond with the amount of profit each subject’s company made. 

Yet recruiters should always be aware of its unreliability. "Intuition is brilliant in predictable situations – that is, where you face the same choices with the same variables again and again, for example firefighters, bakers, people driving a car. Tapping into expertise in situations like this serves us well," says Jonny Gifford, senior research advisor at the CIPD. "But assessing individuals involves a complex mix of variables and there is so much potential for unwanted bias, you have to root it out as much as you can." 

Would you trust hiring decisions to AI?

Technology, inevitably, is viewed as part of the answer. Dozens of start-ups are jostling with existing suppliers to offer AI-enhanced selection tools, which use gamified psychometric tests or highly refined algorithms to balance diversity requirements with quantifiable data on individual performance. 

In general, they can demonstrate qualified success in streamlining processes and reducing time to hire. Evidence of improved long-term performance of hires is inevitably trickier. And technology is not infallible. Tech giant Amazon was forced to abandon one of its first AI recruitment platforms in October 2018, after it emerged the tool was favouring male candidates and penalising applications with the word ‘women’ in the mix; its algorithms had developed measures of success from CVs that, in line with the tech sector, came from a majority male application pool. 

"One huge issue with algorithmic bias is that question: ‘who is coding the algorithms?’ The answer is primarily young men," Swart says. "If the coders or the people creating these recruitment platforms have biases, there’s a real risk the technology will intrinsically take on the biases people had in the first place. And unfortunately, when left unchecked, algorithms and AI will learn to pick up those biases over time." 

Kate Glazebrook, behavioural psychologist and co-founder of blind recruitment platform Applied, argues recruitment should never be entirely given over to the robots. "A lot of technologies take existing data and use it to predict who should be interviewed or given a job, because to build those algorithms you must rely on existing data – which contains bias," she says. "It’s important that technology augments and enhances the way everyday people make decisions, rather than replacing it." 

For Swart, who helps organisations in high-stress sectors choose the best hires, building practical neuroscience into recruitment is crucial. "When you are confronted with two equally good CVs based on qualifications and experience, the single factor that will influence someone being successful and sustainable in that role is mental resilience," she says. "When the going gets tough, the results are bad and people aren’t getting on, you need someone who will still be able to perform, get along, and work well for the company." 

Candidates hoping to work with Swart’s clients undergo a rigorous neuroassessment process to gauge their suitability; they wear monitors to assess their heart rate variability, undergo blood tests and keep food and drink diaries, and take part in a detailed psychometric assessment. These profiles, Swart says, will measure their resilience, but also goal directedness, strength in abstract thinking and sense of belonging. 

"It’s about ensuring the right mutual fit. If a person, for example, needs to settle down and get a salaried job for personal reasons, but accepts an offer because it is the first one that comes along, that won’t be sustainable for them," she says.

This is, perhaps, the apex of luxury recruitment processes and is inevitably impractical for most hires. But even if you can’t run blood tests on your applicants, there are still practical actions that can disrupt the ingrained biases of recruiters. 

Think hard about what you’re asking recruits for

This begins, the experts agree, with a well-crafted job application. Failing to use inclusive language and avoid gendered terms could kill interest from up to half your talent pool before you have even begun. It’s a broadly cited statistic that men apply for a job when they meet only 60 per cent of the qualifications, whereas women need 100 per cent of them; software company Atlassian reported an 80 per cent increase in the global hiring of women in technical roles over a two-year period after it used an augmented platform to analyse the language in its job adverts.

Recruiters should also know exactly what they are looking for, and be selective in what they put out. "Go back to basics and do some proper workforce planning – you need to have a laser focus on the needs of the job and not become too sidetracked by the ‘nice to haves’, because it can be easy to have a long list – but if the number of requirements are too great, people will prioritise based on their preference," says Gifford.  

"For some jobs, you will need brilliant emotional intelligence and presentation skills – if you work with customers face to face or present pitches, for example – but for a back office processing role those requirements should not be there, because they are not a core attribute of the role." 

Glazebrook advises that applications are randomised and assessed on a question-by-question basis, which helps to disrupt problems with successive contrasting bias and offers a more balanced view of the responses offered. Chamorro-Premuzic, meanwhile, recommends performance-led interviews that drive managers to focus on skills over personality. "The most promising recruitment technology enables us to not necessarily ‘replace’ but sterilise and optimise the interview process," he says.

Here, digital tools offer another solution in the form of online platforms where a candidate answers skills-based questions linked to a relevant competency framework. But this brings another caveat: where recruiters are expected to watch video clips of candidates, bias re-enters the process. "The very reason people love the interview is that it gives you things you should actually ignore," Chamorro-Premuzic says. 

"The only way to ignore those things is using machine learning – you need good performance data and models that are theoretically meaningful. It takes work, but with the data that companies are gathering and smart people are analysing, it’s the way forward." 

Arguably even more important is to ensure a diverse group of decision-makers are involved at every stage of the process. "One of the things that has helped organisations change is including someone who plays the ‘conscience’ in the room, ensuring different people interview or assess candidates so hiring managers are not homogenous, and taking steps to ensure no one sees another colleague’s assessment until the end of the process," says Jan Hills, founder of Head Heart + Brain. "There is not enough challenge to recruitment at the moment – and HR processes are not being backed up by the scientific evidence of what works." 

There is certainly logic behind this principle. Almost every employee will have a tale of being marginalised or overlooked by a manager, often for opaque reasons. And it happens to the best of us: consider the high school basketball coach who dropped a young Michael Jordan, or the producer at RKO Radio Pictures, who inadvertently became the stuff of Hollywood legend when he wrote in a 1930s screen test report: "Can’t sing. Can’t act. Balding. Can dance a little". His subject was Fred Astaire.

Research from Applied found the biggest bump in hiring accuracy occurs when three different people are involved in recruitment decisions; dropping from a one in three chance of picking the wrong person to around 15 per cent. For the best odds, decision-makers should be selected from across the organisation and submit views on a candidate without conferring, to avoid falling into collective biases or groupthink. 

"You get the best range of responses by averaging individuals rather than weighting them differently, so we don’t give hiring managers any additional influence," Glazebrook says. "Of course, their perspective is hugely valuable, but the reasons you have other people is that they might spot things that the hiring manager will not." 

All these processes could be rendered redundant, however, if an employer is not clear on what it is looking for. "Organisations have got to think carefully about what they are trying to achieve with a hire: are they looking for the right qualities, have they worked out exactly what the job entails before trying to fix it, what might be suitable for them in the future?" John Hackston, head of thought leadership at The Myers-Briggs Company, says. 

"Companies shouldn’t just be doing this once every 10, five or even two years, but all the time – because what was necessary for your organisation in the past may not be relevant now." 

Look at the big picture

When considering relevance, and taking a step back from the clinical solutions to poor hiring, you could argue the overall structure of recruitment needs a holistic reinvention. The workforce is currently as non-linear and changeable as it has ever been, with portfolio careers, the rise of agency and gig economy work, and developments in remote roles; all of which needs to be reflected in the way organisations take on new workers. 

"Organisations know the majority of employees are not future-ready and may be in jobs that might be automated or outdated soon, and they have a responsibility to cater to that employability in the longer term," Chamorro-Premuzic says. 

Work itself is also increasingly non-linear. The CIPD’s 2018 Working Lives Report found seven per cent of people now work on atypical contracts, whether part time, zero hours or reduced hours. According to the department for Business, Energy and Industrial Strategy, almost five per cent of the UK population worked in the gig economy over the last year – equating to 2.8 million people. "More people than ever now supplement full-time employment with other activities, and they are becoming one of the biggest chunks of the workforce," Chamorro-Premuzic says. 

This poses profound macroeconomic questions, but it also ought to exercise employers. Simply seeking a full-time replacement for each departing worker is arguably a reductive way of viewing recruitment in what is a fluid economy: flexible working and job sharing may be more profitable ways to allocate work, while in 2017, a global EY study found 50 per cent of businesses were increasing their reliance on contingent workers.   

And then there is the power of looking at your existing workforce afresh. Is it really more efficient to return to the market when you need a new hire, rather than examining the potential to upskill existing employees? 

"There’s an interesting connection that has not existed before between hiring and training," Chamorro-Premuzic says. "If you hire the right people with the right potential, they will be trainable, even if you don’t yet know what you are training them in – whether it’s AI, data science or security. Greater numbers of people are likely to benefit from that training if you can hire well – and manage to keep them." The recruitment model of the future doesn’t come with any easy answers – but if one thing’s for certain, it’s that there’s no excuse for perpetuating the status quo.

This article was first published in Management Today’s sister title People Management

Image credit: David Becker/Getty Images



Find this article useful?

Get more great articles like this in your inbox every lunchtime

Why is Primark finally going online?

Primark has finally joined the digital sphere with a trial click and collect service -...

Inside Latvia's first unicorn

Print-on-demand group Printful have become the first Latvian start-up to gain unicorn status. MT speaks...

What radio can teach leaders about the metaverse

"TV didn't kill radio. The Metaverse won't replace reality," says the CEO of ad agency...

Managers who are honest about failure make better leaders

Podcaster and author Elizabeth Day urges organisations to be more open about mistakes

“You are not going to get better by accident”

5 minutes with… Rachel Cook, managing director at digital design agency Thompson, who rose through...