Smarter HR Decisions through Big Data Analytics

laptop

Man versus Machine

Money might make the world go around, but data tells it how fast and in what direction to spin, and 91% of top executives acknowledge this.

Using HR analytics means learning from insurance companies and credit scoring agencies to combine large sets of anonymous data that predict an individual’s behavior in a similar situation.

Big data, with its three key features (volume, velocity, and variety) is the foundation of modern analytical systems. Financial and marketing departments have already acknowledged and leveraged the power of machine learning algorithms to identify profitable stocks and cater to clients in a customized way.

For HR purposes, Google, a data-driven company, has developed a model that predicted promotions with a 90% accuracy. Managers do not use the equation since people are not ready to hand over control to a computer. It is not a failure, but more an indicator of the perception of such models in a real-world setting at this moment.

Big Data for HR Purposes

Using a statistical model means replacing presumptions with validation, hunches with data, and intuition with success ratios, rather than abandoning responsibility to an algorithm. Talent scouts face a significant challenge of structuring large quantities of different information in a way that can be evaluated against a unitary scorecard. Big data offers just that by employing methods specially designed for the 3rd V (variety).

Decrease Cost of Bad Hires

What can an HR analytics model give an organization? The short answer is: cutting down on bad hires.  A wrongfully chosen employee costs the company much more than their salary and benefits. Recruitment costs, training expenditures, productivity loss and negative reviews from clients are just a few examples of what the wrong man in the wrong place can do. Using big data to predict a match between the candidate’s skills and personal beliefs against the company’s needs and driving values is the primary challenge of HR analytics.

Increase Retentions Rates

Who is going to leave? The algorithms created by big data consultants nominate individuals by studying the employees’ online activity, profile updates, employment history, job performance and payroll data. Therefore, if the computer red-flags an employee, it is time for a raise, a more challenging role or some more training. Major companies, including Xerox, Wal-Mart and Credit Suisse already do this with impressive results, most noteworthy, increasing retention up to 20%.

Keeping skilled employees is a long-term goal, but avoiding toxic ones should also be on the list. Cornerstone analyzed a quarter million observations and determined the characteristics of negative employees: unreliable and self-proclaimed rule followers.

Predict Performance

What is the best profile of a candidate considering the job’s requirements and existing top performers? HR analytics models use existing records of successful candidates to create profiles of high performers automatically. The aim is to design a targeted head-hunting tool able to send personalized messages to the right talent. Some freelancing platforms already use this approach to recommend candidates by combining past success rates at similar jobs and current job requirements.

Predictions are necessary to evaluate future job openings, promotions, and even layoffs in a company. Consequently, by aligning models to the company’s business strategy, big data could help save time and money on recruitment. The US Special Forces are already doing this and have selected grit and the number of push-ups as the leading indicators. Each company can define their own, and small businesses can use the aggregated data of similar entities.

Improve Benefits Packages

What perks do employees want as part of their salary package? Following the insurance companies’ footsteps, employers gather health related data of their staff and candidates. The results should only be used to create more attractive and useful packages. Also, the company should be transparent about collecting such data stating the final goals, to avoid legal issues related to discrimination practices.

Legal & ethical issues

Privacy is an important concern when big data is the subject. Most people are afraid that the numbers could work against them and report such practices as discrimination. However, under the anonymity protection of large volumes, it becomes just a risk management technique.

Although there are no legal statements against using big data for evaluating HR, ethical concerns stop the process of mass-adoption. Stereotyping and unfair treatment of an individual outlier based on the general performance of a group is an unsolved problem. Algorithms do not have intuition and are unable to assess undocumented progress, such as that made by a self taught person.

Organizational maturity

The ability of an organization to make the most out of automated models resides in its process maturity level. Big data is just another tool to streamline operations and increase productivity.

Josh Bersin, a founder and principal at Bersin by Deloitte, proposes a 4-step model of talent analytics depending on maturity. The first stage is the reactive approach focused on ad-hoc reports and firefighting. Secondly, comes the proactive style which uses industry benchmarking. The superior levels rely on statistics and data analytics. The third level uses historical data to understand cause and effect while the fourth makes predictions and uses scenario planning. The fifth level of complete automation is currently utopic. Most companies have not even reached the third step, and just a handful are accustomed to predictive methods.

Conclusions

The same algorithms that are helping us decide our next purchase on Amazon could help employers select the best of us from the candidates’ shelf. Our education, experience, know-how, and network are becoming product features. However, we are not living in the matrix. For now, it is safe to assume that algorithms will be used just for screening purposes. Interviews, tests and personal interaction are still the primary recruiting tools, but analytics ensure that less promising candidates do not get an invitation, saving time and money.

, , ,