Recruiting with AI: Winning the Race for Talent
Table of Contents
Alongside the digital transformation over the past decade, developers have been working on Artificial Intelligence (AI) to help us with time consuming tasks, save us time and improve our daily lives. AI now helps us get to our destinations faster and predicts the weather better. These self-learning machines can analyze vast amounts of data in milliseconds and provide insights that make us smarter, more efficient, and better at the things we do every day. This is a key reason why AI is one of the hottest buzzwords in the business world today as executives search for ways to not only become more efficient, but also better at what they do. In talent acquisition, recruiting tools like applicant tracking systems have been able to automate mundane processes to help save time, they haven’t necessarily been able to help HR teams work smarter. Simply put, these tools lack the ability adapt on their own and provide the insights HR professionals desperately need. According to a recent survey by recruitment firm Hays, 92% of employers surveyed were seeing skills shortages that slowed their hiring and negatively affected their business. While AI can’t magically give candidates skills to fill those gaps, it can help identify and automatically target more relevant candidates that are the closest fit. And that optimism is catching on; 80% of executives believe that AI can help make their recruiting process more efficient.
To help you better understand the role AI can play in your own recruiting process, we’ve created this guide to illustrate how AI recruitment can lead to more effective candidate sourcing, screening, hiring, and retention.
What is AI for Recruiting?
One of the biggest challenges facing HR professionals today is finding the best talent to hire. This task has proven burdensome in the past, in part due to inefficient manual tasks that plague the recruiting process and the lack of access to the right data to make informed decisions. In fact, per research conducted by LinkedIn, 46% of recruiters and hiring managers have identified “finding the right candidate” as the biggest hurdle in hiring today.
To tackle these challenges, new technology companies are rapidly emerging in the HR tech ecosystem with robust solutions that use Big Data, predictive analytics, and AI to automate and improve everything in your recruitment process from job advertising and resume screening to applicant engagement, scheduling, and recruiting by text. These new tools offer us ways to help overcome the limitations and biases inherent in recruiting with automated processes that are hyper-responsive to market data, complex metrics, and even budget constraints.
To truly understand what AI recruiting is all about you need to first decipher the techno buzzwords that come along with the hype. Let’s start with the basics and work our way up to AI.
Automation mimics human rules. It saves time and eliminates the manual effort required for time consuming tasks, but it doesn’t necessarily make the recruiting process any better, nor can it adapt on its own. In some cases, it causes more harm than good. Just think about all those qualified applicants the ATS accidentally kicked out due to a mis-match of words. In addition, the pre-defined rules that control the basic automation do not change until the human goes in and manually modifies the code so improvements come few and far in between.
Big Data is not AI—but without it, AI would not exist. Big Data is the process of aggregating, analyzing, and correlating vast amounts of data from disparate sources with the goal of finding insights that may not be so obvious on the surface. These powerful insights translate into better decisions made by humans and machines.
Predictive Analytics is all about finding patterns in Big Data that may have impacted past outcomes such as job ad performance or applicant engagement, so we can better predict future outcomes. Once the patterns or attributes are determined, predictive models can be built that enhance the decision-making process and can transform hard-coded rules into “adaptive logic” that produce better results. Think of it like an “educated guess” we humans make, but much more accurate and dependable.
Machine Learning is not AI, but it is a subset of AI that trains machines to learn by automatically applying complex mathematical calculations to Big Data over and over again to find those patterns without human intervention. From there, predictive models can be built and refined automatically over time as the machine learns with each iteration’s findings and as new data is introduced over time.
In a broad definition, AI enables computers to do things that, without it, would require human intervention like complex decision-making, problem solving and learning. This technology not only significantly improves the process and outcomes, it also takes into account things that are not planned for or known by humans, using data to make the best decisions when it comes to carrying out a task. Self-learning algorithms are required when the scope of data and the scale of the problem are just too big for human interaction.
Individually, any one of these technologies is useful, but when they are all combined together, they become very powerful. This is what AI recruiting is all about. It’s about helping companies improve the way they recruit talent!
What AI recruiting is not: a magic program that eliminates the need for human decision-making or knowledge. Rather, the AI programs that help maximize outreach and candidate evaluations put the human gatekeepers in a better, more informed position to hire the right people for the available jobs.
How Does AI Impact Your KPIs?
Because AI requires data, data, and oh, more data to work, a place where you’re likely to see the most immediate benefit with adoption of AI-enabled recruiting solutions is in your key performance indicators (KPIs) like time to hire and cost per hire. These areas are probably where you are feeling the most pressure as well. Not only can AI help improve your day-day operations, which will have direct impact on your costs, it also offers new insights that will help you improve your overall strategy. You may even discover better KPIs to use along the way.
Cost Per Hire
According to SHRM, the average cost-per-hire (CPH) is $4,129 per open position. This statistic keeps rising year over year as companies are forced to spend more on things such as posting on job boards and employment branding to fill their talent pipelines. In addition, HR departments are having to increase their staff to support all the extra work related to sourcing and screening to make the right hires in a timely manner. The only way to keep cost-per-hire in check without adversely affecting time-to-hire is to reduce costs or improve efficiency, and both are exactly what AI-enabled solutions aim to do. These data-driven solutions are capable of targeting the right job seekers and deliver better quality applicants faster at lower costs. They can screen candidates more comprehensively and for particular qualities without human intervention. They can even engage with candidates instantly no matter what time of day, so you don’t let the best ones get away just because your ‘out of office’ light is on. The streamlined processes and superior results that AI delivers will definitely impact cost-per-hire in a positive way.
AI-enabled recruitment technologies like pandoIQ, a programmatic job advertising platform, are using predictive-analytics as well to work smarter and deliver better results that lower cost-per-hire. Why is it so important to job advertising? It’s one of the biggest expenses in the cost-per-hire equation. In fact, according to a recent LinkedIn study, the average company spends 30% of their recruitment budget on advertising on job sites. Unfortunately, it’s also an area full of haphazard spending due to the lack of transparent data to answer when and where to post jobs.
With access to the Big Data, AI-enabled algorithms like those found in pandoIQ can predict performance in advance and answer tough questions like “how much spend does each job ad really need?”. With the predictive data at hand, these algorithms can identify those hard-to-fill jobs that will require more spend and allocate your job advertising budget accordingly. You end up with the best performance possible from your budget across all your jobs that need filling, without having to spend more. That is definitely good news for your cost per hire KPI.
For example, pandoIQ predicts Job X will get a low number of views and applies compared to Job Y, indicating Job X is harder to fill. It will then re-allocate some of Job Y ‘s budget to Job X which needs additional spend to get the desired number of applicants. This also ensures that those easy-to-fill jobs that get lots of clicks don’t run away with your entire budget as well.
Time to Hire
When it comes to decreasing time to hire, automating the recruiting process is one way to fill job openings faster. Unfortunately, automation alone, which is based on pre-set rules determined by humans, does not have the power to improve outcomes. It simply speeds up the process in the same way that automation has sped things up since the Industrial Revolution: replacing human tasks with ultra-efficient processes so things can be produced faster, but not necessarily better.
One example of how AI can help reduce time to hire where automation alone is not sufficient lies within the ATS. Applicant tracking systems have done a great job at automating straightforward tasks such as scheduling, providing status updates, triggering workflows, and more to speed up the hiring process. But, when it comes to more complex processes like resume screening, these solutions have had adverse effects on time to hire. According to CIO.com, over 75 percent of applicants are automatically eliminated in the hiring process by the ATS and 62 percent of companies using applicant tracking systems admit that some qualified candidates are likely being automatically filtered out of the vetting process by mistake. One of the main reasons of the accidental elimination is due to the ATS’s inability to read the resume contextually like a human would such as taking into account synonyms that mean the same thing. Instead, the ATS uses hard-coded screening criteria specified in the job requisition to do an exact-match with text from the applicant’s resume. So, if the recruiter specified that a resume must contain the job title “Territory Sales Manager” in the screening criteria, but a highly-qualified candidate used a synonym such as “Field Sales Manager” instead, the application could easily be rejected since there was not an exact keyword match. Accidental elimination of applicants meant companies needed to spend more time searching for the right hire which was incredibly time consuming.
Now, AI-based applicant screening solutions tap into natural language processing and comprehensive job taxonomies to solve the problem automation alone could not. AI gives the machines the power to learn, think, and analyze vast amounts of unstructured data found in text using complex human-like logic resulting in screening accuracy rates exceeding 94 percent. Not only does AI save time, but also makes ‘informed’ decisions that keep more qualified applicants in the funnel ultimately decreasing time to hire.
Another example of how AI can help cut down the time-to-hire is by providing predictive data and timely insights that can help you optimize your overall recruitment strategy. For example, you can use the predictive-performance data provided by job advertising solutions like pandoIQ to prioritize how and when you assign additional recruitment resources to fill a job. If the job ad will perform well, then there is no need to tie up more costly resources like recruiters to manually source applicants. Instead, they can focus their valuable time from the get go on the hard-to-fill ones where the job advertising alone may not be sufficient.
Improving Hiring Quality With AI
Improving Quality of Hire
While time and cost are certainly major considerations, improving quality of hire (QoH) is still the king. More than 40% of recruiters and hiring managers cite it is the most important factor in hiring. To put a number to the QoH (also known as the QoH Index), you can attach a percentage to the following factors:
- Factor 1: New hires’ productivity levels (which may include revenue or sales or meeting project goals within certain timeframes)
- Factor 2: Surveys both for new hires and hiring managers assessing performance (creating data using a scoring metric)
- Factor 3: Retention rates and turnover
From there, each factor can be calculated as a percentage, added, then averaged:
QoH Index = (Factor 1 + Factor 2 + Factor 3)/ 3
But how do you get the answer when you don’t even have the data within reach? Luckily, certain new solutions are capable of capturing decentralized data from all kinds of sources, ranging from simple internal satisfaction surveys to comprehensive performance evaluations. From there, they use Machine Learning to find the patterns in the data that can deliver the answers you need—a holistic picture of the QoH.
Reducing Hiring Bias
The data collected and reported by AI-enabled recruiting software doesn’t just help study and improve the QoH metric; it also helps eliminate some of the most problematic human flaws in the hiring process. Eliminating hiring bias is one of the biggest challenges in recruiting today. Whether explicit or implicit, stereotypes and personal biases are something to which even the most conscientious recruiters can fall prey. AI helps level the playing field by allowing for blind applicant screening. For example, Textio is a web-based tool that checks job descriptions for words that indicate gender bias. Other recruiting and applicant tracking software programs can also help generate interview questions that are free of bias or other human nature factors.
Reaching Qualified Passive Candidates
Passive candidates, or candidates who aren’t actively seeking a new job, can actually be some of the best applicant pools. Reaching these candidates in the past meant a lot of resume database searching, cold-calling, outreach, and dead ends.
Although real-time matching technology can eliminate the hassles of keyword searching to identify passive candidates to target, some new solutions like Entelo are taking it one step further by aggregating a candidate’s profile information and recent activity from different sources, such as public resume databases and social networks, to predict how “receptive” the candidate may be on taking a new job as well.
Another way to target passive candidates is on social networks. In fact, according to Jobvite, 67% of people who found their most recent job on social media used Facebook as part of their search. Unfortunately, targeting on social media sites requires building the right audience profile for each one of your jobs. Ads must be tested and audience targeting must be continuously refined over time. Programmatic job advertising platforms make the process easy and can even optimize and learn over time so your job opportunities get in front of the right passive candidates instead of the ones you don’t need.
How AI Enables Audience Targeting
The “purple squirrel” is always the goal of any hiring process. The data culled, developed, and reported by software programs should bolster your strategic sourcing initiatives by providing real-time information such as which job sites perform the best based on the job type, how many candidates fall off in the process and where, and so on. The stronger the data set is, the better you should be able to target your recruitment resources to particular sites or channels that are likely to yield big candidate results. Sounds logical, right? Maybe if you’re a data scientist and you have all the historical information you need neatly centralized in one place. But, that is usually not the case for most HR teams which are left in the dark making educated guesses on things like when, where, and how to get the word out about their open positions.
This is where Big Data and Machine Learning come to the rescue to analyze the data and automatically focus your job board spend toward the highest-yield channels. For example, if you got a lukewarm response from Job Board X the last time you hired for this position, but had many interview candidates from Niche Job Board Y, wouldn’t you allocate more of your resources more toward Niche Job Board Y next time? If it were only that easy with so many job sites to choose from across all your different types of jobs. By tapping into Big Data and letting the AI guide the process, you’re setting yourself up for a better ROI—not to mention a better candidate pool—without having to do the guesswork.
Programmatic job advertising platforms like pandoIQ use AI-enabled algorithms to mine years and years of historical performance data across millions of job ad campaigns to determine the ideal targeting strategy for each job type. These sophisticated algorithms are capable doing much more than just determining the best places to advertise your jobs. They also calculate the right CPC bid rate site-by-site, and even determine how long a job ad should be promoted on a specific site before it will no longer get results.
Why Big Data Analytics is Key to AI in Recruiting
Think of the old-school consumer advertising model. How do companies know how to target ads and messaging to particular demographics? How do they even know which demographics to target to begin with? The answer is simple—research. Since the dawn of marketing time (even back before the Don Drapers roamed Madison Avenue), companies have conducted market research to understand who is buying their products and why. Without sufficient data, they could never determine who their ideal audience is and the best way to reach them. The approach in online recruiting isn’t much different, but it first requires access to tons of data, a.k.a. Big Data. The good news for recruitment when it comes to analyzing all this data is that compared to the days of Don Draper, technology has come a long way. We now have access to tools like Machine Learning and AI that can mine billions of data points in supersonic speeds to identify key trends and attributes that impact the outcome such as ad performance. With access to Big Data, predictive models and algorithms can be created that are capable of automating complex processes and making informed decisions that used to require human thought. We all know things change over time, and so does Big Data. The Big Data model is continuously updated and refreshed with new data from many different sources, which enables self-learning algorithms to constantly monitor changes and automatically adapt over time.
Corralling the various data points you have from candidates and combining it with historical data from employees, the company, the industry, etc., yields a very powerful picture.
The whole concept of Predictive Analytics in recruitment would also not be possible without Big Data. A prime example of this is the NAVi algorithm in pandoIQ, which takes a massive database of past job ad performance information and analyzes more than 199 billion data points across 5.4 million historical job ads using Machine Learning to identify the key attributes that can be used to predict job ad performance. From there, automated algorithms and recruiters alike can use predictive models built off Big Data Analytics to make more informed decisions about what to do next.
Choosing the Right AI Recruiting Technology for You
Because every HR department has different needs, there’s no one-size-fits-all solution for implementing AI. The first step is thinking about your department’s goals, noting the pain points you’re hoping to fix. Then, you can identify the right recruitment technology solutions for your company. The following are a few suggestions to help you get started.
Candidate Classification and Targeting
Do you have a pattern of posting jobs, only to find later that you are not getting what you want? Your job ads are probably being read by the wrong crowd. The best place to start is looking for recruiting solutions that can correctly classify your job ads based on all your specific requirements (not just the title) so the right audience can be targeted. Look for software programs that feature natural language processing and job taxonomies that can convert your entire job description text into a data model that can be used to target your ad the right candidates, on the right sites, at the right time.
Do you and your staff spend a lot of time manually managing job ads across many different job sites when you could be focusing on other more important tasks? Is your team under extreme pressure to deliver more, but spend less? If either of these apply, it sounds like you should be looking for a programmatic job advertising platform. The ideal solution will feature predictive algorithms that automate the entire job advertising process across a vast network of sites and optimize campaign strategies in real-time to deliver better performance and ROI.
Data and Reporting
Are you muddling through Excel spreadsheets trying to get an aggregate view of performance and cost for all jobs? Is your data locked up in silos preventing you from measuring true ROI? Solutions that offer real-time reporting and predictive insights will fix your data woes and help you measure effectiveness from beginning to end. These solutions offer a new level of data transparency that allows you to easily measure ROI across all of your recruitment spend, set better expectations internally, and even improve how and when you allocate your sourcing resources to fill your open positions fast.
Do you have chaotic communications between candidates and departments eating up valuable time and resources, as candidate information pings back and forth between HR, the candidate, and hiring manager? If so, you should consider communication tools that help automate and smooth communication and timing among different stakeholders, as well as analytics and reporting systems that work across other human capital systems to help eliminate data silos and give everyone the 360 view you need.
Will AI Replace HR Recruiters?
There’s a chance that AI will replace all of our jobs some day, regardless of industry or job function. But if you’re worried about AI phasing out the “human” part of human resources, that day is not nigh. According to Deloitte, 38% of companies are thinking of a digital HR world, but only about 9% of them are remotely ready for that possibility.
Instead of worrying about the replacement possibility, it’s more productive to focus on how AI can make things better today. Tasks that would have taken up time or been delegated to junior staff, can be handed off and streamlined by AI, freeing up resources to do other things. There’s also a level of discretion that AI just doesn’t have. A software program can crunch numbers and come up with predictive models, but it doesn’t necessarily have the ability to gauge soft skills or context.
There’s also the applicant experience to consider as well. An automated test or form is not going to be able to serve as an ambassador to the company or answer more qualitative questions about the job or company. There’s still something to be said for the personal touch. Ultimately, you’re people hiring people, and there needs to be a human interaction to help ensure that all is as it seems—that “good on paper” translates to just plain “good.”
AI is Great for Recruitment
If you’re thinking about upping your AI game for your recruitment process, it’s important to understand how it can supplement (or improve) the other processes you have in place. Having AI software won’t magically resolve the traditional challenges of hiring (like finding quality people). But what it can do is help you make more informed, data-driven decisions, as well as streamline the time consuming tasks that can take resources away from talent management or strategy. AI recruiting can make your life easier and it can make recommendations, but at the end of the day, you are the one with the power. AI recruiting just helps you wield that power more efficiently and, ideally, at lower cost.