Recruitment: filtering out the false negatives

The thing that stands out most in this article about online job portals is that a 5% response rate to her job applications is really dispiriting for the seasoned strategist. She wonders where the ‘human’ is in HR. This is when she applies for jobs well within her competency.

The recruitment filters aim to filter out false positives: the aspirational candidates no good for the role. They are ruining it for the honest people, but so are the recruiters. False negatives are as bad: recruiters miss out, and in aggregate, they create an environment in which good candidates still  have a big risk of being a false negative, which incentivises them to put less effort into each application.

That means that recruiters have less information; they have set up a system that is forcing candidates not to do any filtering for themselves by picking and choosing their best jobs, but instead trying as many jobs as they can because every one is a lottery and you have to be in it to win it. Recruiters are encouraging those ‘aspirational’ applications they’re trying so hard to avoid.

Dating websites had a similar problem. Men have a much higher demand for connections using them, and women are inundated with hundreds of messages. This means that they cannot read all messages and any men who wrote careful or thoughtful messages were very likely to have those messages missed. The incentive is therefore for them to give up; to be as scattershot as other men and wait until they get a reply before beginning to make an effort.

A step in the right direction is the swiping mechanism of Tinder or Bumble. This makes the initial screening as quick and easy as possible: each side sets up a profile in a standardised template and only when both are interested is it possible to put more effort in. Imagine users of one of these dating apps being asked supplementary questions by many profiles before being allowed to swipe right: it’s a good way to haemorrhage users and prevent matches. Users will simply start to ignore such profiles. Some users will fill in such questions, but they could as easily be game-playing creeps spotting a niche or fools with too much time on their hands as dedicated, well-balanced partners.

If the filtering of job applicants can be done from a standardised form, as with dating websites, that would help everyone. The equivalent of the initial right-swipe might be a check of qualifications and a minimum number of years of experience. As with dating, filtering by keywords is a great way to narrow the pool, but also a great way to exclude good options. Imagine looking for a partner who loves theatre by only accepting profiles with the word ‘theatre’ in them. A few synonyms might help: ‘plays’, ‘show’, ‘performance’ etc. After all, some people might write ‘I like to catch a show…’

It really needs human judgement to check, in a second or two, whether it’s right, because the possible phrases and permutations are too many to predict in a simple computerised search. That needs a basic profile, even shorter than a CV.

That might mean that recruiters can’t get around to looking at everyone, just as daters can’t. Yet people find soulmates through such a mechanism, so companies tempted to argue that this will lead to disastrous outcomes should think carefully. It might be true that our hormones get us addicted to a partner and deceive us into believing in the concept of soulmates. But anyone cynical enough to take that line must acknowledge that the concept of a soulmate employer is at least as much a deception.

By not reviewing some candidates, however, employers must overcome a number of mental blocks. I suspect that at the moment it’s easy to persuade oneself that the current system of filtering for in-demand jobs is fair and rigorous, because some sort of criteria are applied. Leaving out some people is definitely random. But poor criteria with either false positives or false negatives are also random. It’s better to be sure about some and open with the rest than random with everyone. How should we select those to assess more carefully? If no other method seems reasonable, randomly. Deliberate bias is worse than randomness.

I would welcome being told that there were so many candidates that a good panel to interview was found before I was even considered. This would be so much more informative than the usual ‘sorry but not sorry’ message. When combined with genuine messages (still possibly automated) telling me about failures I would then truly be able to judge what roles people were willing to consider me for and apply to more appropriate ones!

Matching is an iterative process and the key is to get through iterations with as little effort from  both sides as possible. Pushing all the effort onto one side might make the other side think everything is working efficiently but it creates problematic incentives. If you’re looking for sensible, intelligent candidates, you probably want people with the wit to recognise and respond to those incentives, not the chaff who don’t. And that’s purely from the self-interested point of view before we consider the possible effects on diversity and privilege of extra challenges that formal education does not address.

The formal structure of CV-assessment followed by interview is itself a relic of a time when there were far fewer candidates. The leap from CV (or even covering letter) to taking time off work to travel to someone else’s offices is a big one. The SARS-CoV-2 outbreak has pushed alternatives such as telephone interviews. This is also a welcome advance that we shouldn’t abandon. Gradual iteration that is as conversation-like as possible is important: if at any point it’s clear that things shouldn’t continue then no more effort than was needed has been committed to discovering this.

Employers and daters alike enjoy being chased and putting no thought or effort into the matching process. This is a short-term solution only; it encourages stalking, exaggeration and unhinged, irrational attitudes. These are people who can be taken advantage of, but who aren’t going to be genuinely exceptional. These are the people the world is picking right now, but we can change that.