The Impact of AI on Applicant Tracking Systems
As a precursor to HireGround’s CPHR BC & Yukon Webinar upcoming on October 4, 2018, we felt that some background information on how artificial intelligence is being used to improve recruitment tasks would be of benefit.
“Artificial intelligence (AI) for recruiting is the application of artificial intelligence, such as the learning or problem-solving that a computer can do, to the recruitment function.”
Let me start by saying that a better term to use from here on in is machine learning. With machine learning you are in essence training a computer to perform a mathematical analysis of, in this case, a number of applicant profiles as compared to the needs as described in a job description or job posting.
The objective of running this analysis on a batch of resumes of course results in the following potential benefits:
- Saving recruiters’ time by automating a high volume task
For any of you that do a good deal of recruiting, it not unusual to see 300 applicants on average to a job opening. If the market is slow and there are few jobs, this number can rise to many hundreds bordering even into the thousands.With the advent of quick apply features that are offered by both Applicant Tracking Systems and job aggregates/boards such as Indeed or LinkedIn, it has become very easy to apply to openings. Therefore this number is expected to increase.As a result, manually screening resumes can be a very tedious task that is by far the most time consuming component of the recruitment process. It often means that a high percentage of those applicants are not qualified, making this task even less gratifying. Some estimates put the average time it takes to screen resumes for a single hire at 23 hours.Therefore, if a computer system can be trained to automate the screening of resumes such that those best matched for a role are scored highest, and those not meeting the requirements are scored lowest, this is a huge asset to any recruiter.A side bonus of using such a tool successfully can equate as well into a reduce time-to-hire.
- An increase in finding the attributes most required for job role
As mentioned above, reviewing a large number of applicant profiles can be tedious and time consuming. This can translate into recruitment teams losing focus and therefore losing a subjective analysis of which applicants stand out over others. Personal biases may also play a part in a lower objective review of the applicant data.Automated algorithms that measure an applicant consistently for each role with a view to many factors such as the candidate’s work experience, skills, education, location, objectives against the role requirements will deliver a consistent measure not prone to interruptions or tired eyes.It can be surmised from above that the overall results will deliver a far better analysis of those that have applied, and hence the likelihood of a hire being more appropriate. Some estimates indicate that turnover is reduced by some 35 per cent, with the cost per screen reduced by an estimated 75 per cent.
So, how is all of this magic accomplished?
Well, to give you a comparison, when a recruiter or someone reviewing a resume is looking at attributes that meet the requirements for a role, they are likely looking at some of the hard facts around the applicant first, such as education or where they have worked. They may not have time to look at a cover letter or to review additional documents to any extent; or they may look at responses to questions first and leave the other components of the application to review later.
Here are a few simple figures to depict the impact of an AI/machine learning assist:
- Average applicants per job: 250,
- Potential time savings screening per job: 23 hours,
- Turnover reduction by 35 per cent,
- Cost per screen reduced by 80 per cent; and
- No disruption to the recruitment workflow
In essence, it may be hard to review all of the details associated with any one applicant profile: resume, cover letter, documents and responses to questions. Computers, on the other hand, have do not trouble combing through copious quantities of data. They do not mind looking at everything all at one time.
In fact, machine learning related to recruitment can review thousands of bits of data and measure the applicants closely aligned to the data needs. For example, HireGround uses 14,000 plus vectors or data pieces to analyze any one applicant. In human terms of analysis, this would be physically impossible or at least pretty exhausting if it were.
Now, I have only mentioned in my discussion above how AI is being used to evaluate applicants for a job role in relation to scoring their attributes to that role. This is not the only use for AI in recruitment. There are other uses as well. Though they are not the focus of our upcoming CPHR BC & Yukon webinar, other tools that are being developed for better assessing candidates for a job role are chatbots, sentiment analysis, digital interviews, tools for writing job ads, and tools that measure a video interview.
Join me on October 4th to explore “The Impact of AI on Applicant Tracking Systems” and I will cover some of the other areas as well that will change the way we recruit in the years to come.
Marilynn Kalman is presenting the The Impact of AI on Applicant Tracking Systems webinar on October 4, 2018. For more information on this and other professional development opportunities, please visit cphrbc.ca.