Machine learning in recruitment: it works for everybody

null

Machine learning is quickly gathering pace in the world of recruitment. The time to implement a system powered by this subset of artificial intelligence (AI) to find candidates is now.

And benefits to both recruiter and candidate are vast. AI-powered recruitment systems reduce hiring bias and prejudice, because they approach a candidate’s suitability based on pure data. Whereas a human recruiter may dwell upon personality, a machine will use the far more reliable concept of science.

Of course, to gauge whether a candidate is a good 'fit’, human interaction remains vital. It’s effectively the balance of both that provides the best results.

Machine learning will take time to generate maximum returns, but some benefits do present themselves from the start. Broadly speaking, we can divide the evolution of machine learning within the recruitment profession into three separate phrases, each exhibiting more sophisticated applications of data than the last:

  • Ranking candidates
  • Ranking search queries
  • Personalised job suggestions and success predictions

In the beginning, the system focuses on you as the recruiter and will help match and filter candidates. As you progress to a fully-functional machine learning-based recruitment system, candidates will find it works for them as well – because if candidates aren't suitable for one role, the system will automatically match them up with another. They're happy, you’re happy: everyone’s happy.

Here's what each phase of machine learning in the recruitment industry entails.

Phase 1: Long-term investment 

The entire concept of machine learning revolves around the system's ability to self-improve. This means the longer you use it, the smarter it will become. In its first stage, it will focus on ranking applicants. This is vital for high volume recruiters.

At its base level the machine learning algorithm will automatically rank the candidates in your database based on how suitable they are for the role you're recruiting for. 

You can specify criteria per role and if the information you have for the candidate matches it, the algorithm will flag them as a candidate. However, while this is extremely useful for a recruiter looking to match someone to a specific role, this doesn't work for the potential candidate, who could be far more suitable for another role.

Phase 2: Your own private Google

Recruiters that advertise their roles on a website have a huge opportunity when it comes to a candidate's search behaviour. 

Currently, a candidate will find a role via searching for relevant keywords, but if they are deemed not suitable for a role, what then? By employing machine learning to the website's own search function, you can make it work for you – as well as the candidate.

It does this by cross-referencing a candidate's profile with their search history. This means that when they are searching for roles, the algorithm recognises their search history and provides a more rounded but relevant list of opportunities based on the roles they've been looking at. Just like Google, the most relevant results to them will be at the top.

This is the point where machine learning becomes unknowingly beneficial to the candidate. When looking for a role, they won't always know exactly what they're searching for, so by automatically presenting them with more options based on their search history, the process for them is streamlined and they can save time having opportunities presented to them faster.

From the recruiter's side, there will be fewer applications from candidates to roles less suitable for them, because their eyes are more frequently on the most relevant positions. 

But with increasing sophistication comes even more rewards for recruiters and candidates alike.

Phase 3: Getting the right candidates to the right place

The dream is for both parties to be as hands off as possible to the point of connecting candidates with suitable roles. The more sophisticated the algorithm gets, the more this is possible.

By this point, machine learning will be able to use the data it's able to access from your database to suggest jobs to candidates automatically. Furthermore, it will also provide them with the likelihood of whether they'll succeed in the role itself. 

This will make the whole process more efficient by reducing lower quality applications and matching candidates with roles faster.

So, what happens when a candidate sees a role and it doesn't look like they'll succeed? Will they be intimidated by repeating the search, or end up searching for positions that are even less suitable? No. Machine learning algorithms can quickly suggest other roles relevant to their data where they are more likely to find success.

The outcome is straightforward: candidates will be able to move on from frequently dealing with rejection to a more hopeful situation in which they are simply referred to another, more suitable role. The days of spray and pray are over. Talent is no longer wasted.

The latter point is vital – especially in the face of skills shortages in key sectors like IT. In the UK, the rate of unemployment sits at just 4 per cent (as of July 2018), but this doesn’t mean that every business has the skills it needs. Rather, it means that competition for talent is only going to get fiercer. Sophisticated applications of machine learning will make it easier for recruiters and employers to discover candidates with the right skills and potential by offering insights and opportunities that are not immediately obvious to humans.

A machine learning experience for recruiters

Over time, machine learning will have a transformative effect on your clients and candidates – and the recruitment industry as a whole.

That doesn't mean your work is done – technology should never be used as a crutch, but rather as an enabler. The recruiter's role as a relationship builder – centred around human-to-human interactions – has been and always will be the most important part of the equation. A machine should never make the hiring decision, but with machine learning's help, it is now up to you to assess a better, more streamlined group of candidates and ensure your clients really do get the best person for every single role. 

Peter Linas, EVP Corporate Development & International, Bullhorn
Image Credit: Alex KNight / Unsplash