Ask any recruiter what their biggest challenges are when it comes to selecting talent and they are likely to mention any one or more of the following; high volumes of applications, time-consuming or inefficient processes, unconscious bias and a somewhat outdated candidate experience. Imagine being able to reduce and even eliminate many of these concerns in one fell swoop? Unilever South Africa recently launched a new graduate application and selection process which does just that. We caught up with Alex Cresswell from Pymetrics, a global company, who developed some of the technology behind the Unilever solution.
A genius combination of Artificial Intelligence and Big Data
Pymetrics was founded by two young ladies, each with a PHD in neuroscience, and who were driven to find effective ways of matching candidates to workplace roles. There are objective ways that can be used to measure behavioural traits and Pymetrics developed a set of twelve games based on well-established neuroscience exercises. The games are 2-3 minutes each in length and provide feedback on up to 77 cognitive and emotional traits. By understanding the behavioural traits of high performers and then using artificial intelligence to look for patterns in over a million data points they can create an algorithm that will match candidates to an “ideal profile”, thereby predicting a high degree of success in roles that have been mapped. Matches are based on behavioural traits but also have a high correlation to culture-fit – so applicants who may not be suited to a role in company A may be redirected to a more suitable role in company B where they are more likely to be successful.
What does a typical intervention, using this technology look like?
I asked Alex to explain what a typical overhaul of recruitment and selection processes using AI and Big Data might look like. Because there is a need to develop high numbers of data points to predict success this approach is, at present, ideally suited to high-volume roles like, for example, graduate roles. The starting point is for clients to identify which roles they would like to profile and the project team then facilitate several conversations and interviews with the client to refine ideal profiles for the identified roles. At least 50 well-performing job incumbents then need to be identified and these individuals play the 12 games over a two-week period. The Data Science team then get to work and, using augmented learning, they spend two to three weeks analysing the data and collating the matching of behavioural traits for ideal performers in the roles being analysed. Once this process has been completed the first filter stage of an application process is ready to be built.
The recent project at Unilever incorporated a complete overhaul of the company’s recruitment and selection process for graduates – using Pymetrics technology in collaboration with HireVue (digital interviewing) and Amberjack (experts in volume recruitment, outsourcing, technology and assessment), an end-to-end solution was designed which has, to date, delivered impressive statistics:
- Time to hire reduced from four months to between two and four weeks
- Over 50 000 hours of candidate time was saved
- Recruitment screening time was reduced by 75%
Benefits to the business
In addition to the kind of time savings achieved in the Unilever example, there are many other benefits to a more streamlined and efficient recruitment and selection process, as follows:
- Reduced attrition due to more ideal matches between candidates and the job and company they are placed in
- Increased candidate quality, as measured by an increase in hire yield from one in three to two in three applicants (i.e. the percentage of applicants being hired at the final (assessment) stage)
- Significant cost elimination due to a shortened and more efficient cycle
- Reduced risk of unconscious bias
- A vastly improved candidate experience – not only in terms of a shortened process but also the positive impression made by a technologically advanced and efficient selection process
- Applicants receive almost immediate feedback across a broad range of cognitive and emotional traits – whether their application is successful or not
According to Alex the South African businesses with which Pymetrics have interacted have a real appetite to embrace this kind of ground-breaking technology. Pymetrics’ current capability is just the tip of the iceberg in terms of what is possible – as the number of roles profiled increases over time and the amount of data available expands it will be possible to develop data to help businesses improve their age, race and gender diversity. Data can also, potentially, be used to maximize the internal mobility of talent within an organisation and the building of more generic models for more specialised roles will become possible (versus high volume roles).
Up until now applicants have had little choice when it comes to jumping through the hoops of a long-winded selection process – as more and more businesses streamline their recruitment cycle candidates will become increasingly discerning as to which employers have embraced the benefits of artificial intelligence.