The who, what, when, and how of big data.
By Jason Taylor
In the past, success at the executive level consisted of great intuition and risk-taking abilities bound with hope and luck. In today’s business environment, intuition and hope are being replaced by ‘big data.’ With technological advances, organizations are now able to collect large amounts of data and turn it into the fuel that drives a predictive decision-making machine. Similar to other strategic parts of the business, employee selection, succession planning and coaching strategies should not rely on hope. Instead they should – and can – leverage big data to predict success.
But wait, big data is a topic for the IT department, right? Yes, the nuts and bolts of gathering and storing data are IT topics. But investors want higher profits. Customers want better products faster. Governments want business conducted within tighter regulations. In the midst of these demands, executives must predict market needs as well as determine who, how, and from where to fulfill them in a profitable way. And they have to be right the first time. It’s people who make this happen.
Big Data Reaps Big Results
Big data is quickly creating a have-versus-have-not market. Organizations that have data and know how to use it to make objective decisions are methodically outperforming the competition. In fact, the McKinsey Global Institute determined that data has swept into every aspect of business. In terms of importance, data now sits alongside both labor and capital as critical factors for business success.
For example, a retail chain operations vice president can know which specific departments and employees consistently perform best during which hours and days to determine where to expand. A call center operations manager can see productivity by employee down to the day and time to see where improvements are needed.
In today’s marketplace, successful organizations are searching for new ways to collect and analyze the most strategic business data that provides a competitive advantage. The human capital field has embraced the use of big data to identify, select and develop a cutting-edge workforce that will produce and contribute more to the bottom-line at the job level.
People are the greatest expense on balance sheet, so what executive wouldn’t want to maximize their effectiveness? All they need is a systemic way to measure performance and tie it back to selection. Organizations now have the ability to use big data to ensure that every new hire is selected based on their probability for high performance. Good data reduces the intuition and guesswork. Companies can calibrate employee selection criteria, how they measure employee performance and use the new knowledge to improve performance throughout the organization.
In order to leverage big data in an organization, the organization must first have employee data. Not just any data, but valuable high-quality data that provides objective information on the performance levels of each individual in a given role. As experience shows, what isn’t measured won’t improve. The same thought applies when effectively leveraging big data. Measuring data will enable valuable predictions.
Most large companies don’t collect data at the department level, but instead focus on the regional or overall company performance level. When numbers do not meet expectations, organizations make cuts, divisions are sold, or other large-scale group changes are executed. Decisions are typically reactive and executives reach conclusions from aggregate level data. They rarely focus attention on increasing performance at the individual level. Instead, they use group solutions to resolve individual performance problems. Attacking the problems at the wrong level creates a cycle where executives are forced to make decisions reactively with very little accurate information.
With data collected and analyzed all the way to the individual level, big data truly becomes the competitive advantage. Big data helps align company strategies with individual employees’ goals, and helps organizations reward performance based on the most strategic measures. Organizations can hire and use their people as a competitive advantage to achieve priority company strategies.
What to Collect
Big data, by design, breaks the guessing game cycle. Regardless of the job, starting with performance and behavioral data at the individual level rather than at the location, role or other level to create prediction models will increase the probability of hiring employees that will improve profits and company performance. Eventually stock price will also increase.
Performance data is the focus of big data. To be useful, it must represent actual individual daily performance on the job. And there must be a vehicle to translate the performance data to those that have never done the job in the organization. This is where individual behavioral data comes in. Think of it in these terms. A future employee’s performance cannot be measured on the job in areas such as sales per hour, average handle time or percent achievement of management objectives. Therefore, a vehicle is needed to translate the productivity levels (performance data) into a usable format prior to the person being employed in the position.
Behavior provides the vehicle necessary for translating performance into information anyone can capture and compare against on-the-job performance potential. Performance data plus behavioral data can become a powerful combination in leveraging big data to create the probability model necessary for predicting which candidates have the highest likelihood of producing at higher levels.
When to Collect Performance Data
Frequency and timing is critical to effective data analysis. Collect periodic data that provides insight into performance over time. The more granular the information, the more interpretation and insight can be gained from the data. Consider a sales example. A company that collects individual total sales for each calendar year limits the insight and contribution to a prediction model. The company could gain more information by collecting individual performance data at quarterly, monthly or weekly increments. In this example, the company would now be able to evaluate individual-level performance to understand early stage ramp-up times, sales cycles and each employee’s consistency over time.
Allow ample time for a consistent record of performance data to accumulate to avoid misleading conclusions. Perspective (time) can greatly affect how performance data is interrelated. Consider this example regarding training time. A company that collects and evaluates performance from individuals in their first month of employment may make faulty assumptions about individual employees’ performance levels. Instead, the company should consider the fact that the position requires a three-month training period. Thereby a one-month performance data capture will most certainly lead to false conclusions. When training is complete and the new employee has ample time to ramp-up in the position, data becomes more valuable and meaningful.
How to Collect Performance Data
Accuracy, method, and variability are also vital components of the big data analysis equation. Ideally, organizations will use systems to ensure data is accurate and represents true performance. It is important to remember that data collection is just as critical as the data itself. The data will not reveal relevant insight unless there is an accurate account of performance. If there are no systems in place today, organizations should explore ways to improve measurement of individual performance over time to become a data-centered organization. An organizational focus on performance data will provide the data needed to create prediction models that will improve performance employee by employee.
Data collection systems and technology should be designed to assist with efficiencies and also facilitate the job activities. Data collection systems should not become an obstacle for employees to overcome. The systems must be efficient and accurate to make sure data is representative of actual job performance. In order to achieve efficiency and accuracy, the key is to insert data collection in the middle of the job activity. For example, retail organizations are becoming very effective at utilizing point of sale (POS) systems to track and monitor job performance. Many POS systems are designed to track and provide individual-level data such as sales per hour, dollars per transaction, and many other important individual performance measures.
Variability is an important factor people often overlook. Data will not provide useful information in situations where everyone possesses the same value. For example, I recently worked with a client who felt a quality score metric was very important to measure success. But under close scrutiny, we found the data had no variability. Everyone in the position received the same performance score. The lack of variability caused immediate performance data accuracy concerns. It does not make business sense that all employees perform at the same level. Instead, we should see a variety of performance from very low to very high. In this example, there were no reliable insights to help our prediction model.
Increasing Data Value
Leveraging big data to select candidates with the highest probability of high performance is contingent on the quality of data involved in the process. To gain the most accurate insight from big data, a company must first determine the question, and then ensure the data is focused on providing an accurate answer. For example, a restaurant client desired to be known as a high-service dining brand. The company had a choice of data to either focus on secret diner scores or tip percentage earned by wait staff. The client chose to focus on tip percentage. In their world, it was a more stable metric with less subjectivity. More importantly, the client felt the best indicator of high service levels would be reflected through the customer’s wallet since high service level equals monetary reward for the server.
Another helpful tip is to make sure conclusions are not based on a raw data set. Unclean data may lead to misguided and wrong conclusions. Data should be properly cleaned and analyzed to exclude outdated data, missing values, and address situational factors. For example a retail organization may elect to understand behaviors related to performance based on individual sales across the company. Think of the possible challenges related to comparing sales staff performance in small locations versus large, high-traffic locations. There will be much more opportunity to sell in a higher volume store. Does that mean that all large store sales staff are better and higher producers? Probably not. Instead, the store size may affect the final sales number, so it must be accounted for in the prediction model.
Finally, larger sample sizes will help ensure stable and factual conclusions. Consider a role with 1,000 people where data was collected from only 3. The results or predictions would not be stable. Instead it would be much more reliable to collect data from 400 to 500 people in the role to pick up reliable trends and insights based on data that is statistically significant.
Leveraging big data to draw conclusions is similar to an artist’s drawing. The people represent the paper and the data represents the ink. The more ink there is, the more detailed the drawing will be.
Big data is here to stay. Organizations with the greatest ability to leverage big data will continue to outpace their competition. If an organization is in the middle of leveraging data to proactively embrace the future, then the company will more likely be in the competitive pack. These companies should continue to develop the practice to move into the future. If an organization is staring at a blank piece of paper, then it is time to make a case to improve the data situation. The effort and reward for using big data will have worthwhile payoffs.
Jason Taylor, Ph.D., is chief science officer for PeopleAnswers.