Today, more businesses are embarking on IT staffing and are ready to spend huge budgets on hiring the most appropriate and competent IT personnel. Because competition for human resources is fierce, companies apply non-trivial methods to improve their screening and the quality of new hires. Some pay for ads placement on relevant job boards, while others hire ancillary recruitment staff or 3rd party IT staffing consultants to get better value for money.
As any IT hiring process implies a lot of information to be disclosed and become available (such as candidate's personal data, skills and competences, career growth preferences, etc), machine learning can present a better way to identify worthy candidates more effectively. By creating different data models based on the original input data, companies can make more precise predictions and make informed decisions. Data science allows recruiters to build such predictive models and improve their hiring efforts.
As your investments in human capital increase, it's highly recommended that you apply data analytics to significantly improve your internal HR workflows, facilitate hiring process for a candidate, lower down overhead and increase the overall quality of decision making.
Big Data Use For HR Functions
Although Big Data is a buzzword that's been around for few years now and is a practice employed by many companies in various verticals and niches, very few HR departments are actually aware of the benefits it provides. But when you need to process several thousands of resumes a week, Big Data is the solution that can help optimize recruiters' time and efforts. Thanks to machine learning, data scientists and analysts can ensure a better quality screening process.
When clients turn to us for help in improving their existing HR processes, we always say that IT is the most lucrative niche in terms of ROI attained. Therefore, companies should invest more in building own technological solutions to take advantage of HR data. Analytics can provide benefits in every stage of a staffing lifecycle: screening, vetting and hiring, adaptation, training, talent management, dismissals, etc. Yet, talent acquisition remains the area with the strongest potential for data based business intelligence (BI). Since recruitment is the first step in staffing, investments in this area will apparently affect other HR aspects of each particular organization.
Companies that have already built own Big Data solutions in HR are already on the edge of competition. Google and JetBlue are great examples of how data analytics helps change corporate processes and reach ambitious HR goals. Although these two companies have very little in common, their successful integration of data analytics with HR processes makes them stand out!
Let's review these two brands and how they've revolutionized recruitment with use of Big Data tools.
How Google's HR Takes Advantage of Big Data
Google is believed to be a Big Data pioneer. Google's HR teams involve data scientists to many processes from compensation management to building multicultural relationships. Despite a secret nature of many Google's HR projects, some data leaks suggest the company uses out-of-the-box approaches towards IT hiring.
According to an article in the Harvard Business Review, Google's most famous HR Big Data project is Oxygen that aims to determine the quality of applicants for top management positions based on data analysis. Final stages of personality and skills assessment interviews become data sources and points for Google's HR teams. Originally, Google believed that its engineering teams didn't need any leaders to perform well. However, the analysis of data received from 100+ employees showed a strong necessity for IT leadership on Google's engineering teams and also helped identify top qualities for such managers.
Google's talent acquisition approach proves to be very successful, too. The company used data analytics to reduce the "perfect number" of job interviews from 10 to 5 a day, which helped save thousands of recruiters' man-hours and millions of dollars.
According to Google's People Operations Officer Laszlo Bock, initially the company sought and hired the best of the best IT talent only, following the "Hire those who are better than you!" rule. Nonetheless, as the number of job applicants increased significantly, Google had to process more and more data and began employing data analytics to convert their insights into actionable BI plans.
At present, Google believes each individual job success or failure is defined by 4 key factors: general cognitive abilities, aspiration for situational leadership, non-trivial thinking and skills required to get the job done. Data analytics allowed Google to understand that it doesn't need Ivy League graduates to keep its technology developments at the highest professional level. It made Google change its decision and focus on ambitious and talented specialists regardless of their education, geography and honors.
Also, Google uses its unique machine learning algorithm to revisit previously rejected applicants and find among them talented engineers who might be omitted at the initial screening. It demonstrates that Google makes tremendous efforts to derive value from applicants' data and build effective algorithms to process it.
To date, all of Google's hiring decisions are highly dependent on data analytics.
How JetBlue Airlines Uses Big Data for Staffing
That's clear as a day that not every company can boast the same technological level as Google, nor can every company afford to hire IT specialists able to build Big Data algorithms for HR teams as pet projects. Yet, it doesn't mean that only tech giants are able to capitalize on data they gather, analyze and store. JetBlue Airlines is a good example of how a non-tech company can be ahead of the curve with regards to data insights.
Using data analytics, the company managed to full satisfy its personnel needs. According to Andrew Biga, the director of talent acquisition and assessment, and Ryan Dullaghan, manager of people assessment and analytics, JetBlue always focused on hiring customer friendly flight attendants who can help set out passengers' fears and concerns. Wharton Business School specialists analyzed JetBlue passenger feedback and came up with surprising results: most passengers would prefer to fly with a complaisant rather than a friendly flight attendant. Having re-focused their talent acquisition strategy to seek complaisant rather than friendly staff, JetBlue was able to see a much higher passenger satisfaction rate.
JetBlue and Google used data analysis to breathe new life into their hiring processes, eliminate bias, increase efficiency and, finally, save time and money.
How to Apply Machine Learning To IT Staffing
We recommend you use your talent acquisition team or department as a central hub for data generation and analysis. Recruiters' ultimate goal is to find the right people to fit unfilled roles. Each recruiter has a plethora of connections, but only true professionals know how to make use of them. You can apply machine learning to filter job seekers' resumes without hiring any additional staff. As a result, recruiters save time they'd spend on reading resumes and can focus better on personal communication with applicants.
Machine learning can be used to facilitate a lot of other HR tasks, too. For instance, by analyzing previous job seekers' experience, you can teach your algorithms to identify candidates who'll be most dedicated to your project or who won't pursue any external career opportunities in the mid to long term.
When you let algorithms do your most routine and boring job, you'll see a higher ROI as a result of money and time savings enabled by data analytics.
Besides collecting resumes, you need to put in place a well thought-out process for quality interviews. Most employers keep asking applicants the same questions (both generic and specialized ones). On the one hand, repetitive questions help candidates come better prepared to the interview. On the other hand, it allows for a manageable candidate's assessment monitoring. Having gathered data obtained in the interviews, you can use machine learning to identify candidates that best match your selection criteria.
As mentioned above, machine learning models accelerate hiring decision making. Besides this, a data model created as a result of the interview data analysis helps better evaluate how each job seeker matches a required job role. Using machine learning models, IT recruiters can hold several interviews to determine whether they should hire this or that candidate. After the interview, a recruiter will be able to evaluate each interviewed candidate as a potentially good or a potentially bad one and correct this assessment later after comparing a potentially good candidate with those really good specialists hired before.
Once the algorithm is up and running, you need to fix some errors to teach your model how to distinguish false-negative and false-positive decisions. A false-negative decision is made when a rejected candidate might become a potentially good employee; having rejected the candidature, the recruiter will never find out if this person would have been a good or a bad employee. In this case, you may need to hold additional interviews with rejected candidates to make sure you were right to exclude them form the process. Each new set of data helps better teach your model.
A false-positive decision is made when you shouldn't have proposed a job to a potentially good candidate. Under this category fall candidates that accept job offers, but appear to be slow achievers and bad performers.
Nevertheless, you need to take into consideration expenditures related to corporate decisions imposed by machine learning models. The cost of false-positive decisions is very high. You, as a company, invest your time and money into a person who doesn't have skills or zeal required for the job to be successful. It harms both your product development and your team environment. The cost of false-negative decisions is an overhead the company will have to pay as a result of rejecting a potentially good candidate.
That being said, many recruiters make false-negative decisions hoping that candidates will still apply for company's job openings in the future and, thus, will help them make more accurate predictions.
When applied to HR hiring, machine learning algorithms have own pros and cons. However, they always prove to be more effective than the traditional approaches. In spite of likelihood of false negatives and false positives, a human error poses a much higher risk to integrity of processes.
Although machine learning algorithms won't replace human involvement completely, they'll undoubtedly speed up screening and hiring, and ensure better ROI.
Economic value of machine learning application in HR
A hiring process including recruitment, job promotion, candidate screening and selection always has a hidden agenda. Applying machine learning algorithms requires considerable investments; however, as proved by Google and JetBlue, they pay off well. By using methods described above, you'll be able to optimize your costs and improve your business value.
Let's take a look at the following example. To replace a resigned or a dismissed employee, you'll pay on average $4,000 to $7,000 per person depending on the job role, experience and seniority. Machine learning algorithms could minimize your overhead resulting from making false-negative or false-positive decisions and drive your corporate growth in the mid to long term.