I have my own favorite story on people analytics and this comes from a time when we did not have ‘big data’ systems to manage, hold and understand people-related information. I was managing human resources in the Philippines for an outsourcing company that had around a thousand employees and quite a few demanding customers. We believed we had a serious employee engagement problem, which was being manifested in the extraordinarily high levels of attrition we faced. Indeed, at that time we were running at more than a 40 per cent turnover—the task of recruiting and retaining people was consuming most of our time and mind space. This was proving to be a constant source of worry for all of us. Given that the high attrition was hurting the business so much, we went into an overdrive for engaging employees, using a mix of approaches. At one point, we were even sending roses to our female employees on Valentine’s Day.
Sadly, even after these efforts, the attrition numbers did not move. One day, on an impulse, I decided to relook at the files of the employees who had recently exited the organization. Using an information template, two of my HR managers were tasked with getting as much data on these employees. The analysis of the data revealed some amazing trends—all the people who had left us were either studying nursing, living at least one-and-a-half hours away or were working for an Indian company for the first time. Armed with these insights, we took action on the recruitment side of our business and, sure enough, our early stage attrition (zero to six months), which was our largest bucket, crashed.
My learning through this experience was that an assumptive point of view is a dangerous luxury when it is not backed by data and review. Even when the data such as this exists, sometimes the managerial lens might miss these patterns and connect the dot meaningfully.
Today, technology helps us make sense of the available information and connect disparate data buckets. It is proving to be a valuable tool for us in the HR function. Like other fields, we too have seen an explosion of data that is available to understand, analyze and even predict potential problems. The aim, however, should be to spread the net wide, be wary of any validation bias (using data selectively to validate what you hypothesized) and to understand the context of the data generation process. This has helped us considerably in AkzoNobel India where I am currently employed. We take considerable care to understand the core issue, frame its context, study the data as widely as possible, derive insights and only then initiate action.
One of the key challenges that the Human Resource department of an organization faces is managing attrition. Talent strategy is, in fact, as important as any other part of an organization’s overall strategy. Poorly informed decision and strategy can adversely impact retention rates.
Once, during an attrition review meeting, we were trying to identify and concentrate on functions with high attrition. Instead of following the normal highest percentage method, we encouraged the managers to first map the demographic function curve on top of the attrition curve and then select the functions which were outliers. This changed our direction of effort—here’s the playback—we believed initially, for example, that because our highest attrition per cent was in sales we should target sales for most of our retention actions. After mapping the two curves (demography and attrition), we realized that there was another function that accounted for just two per cent of our demographics but contributed to five per cent of the total attrition pie. Sales, on the other hand, was 50 per cent of our people and contributed to 50 per cent of our attrition. So, while the latter function (sales) was within an acceptable range, there was clearly a problem in the former function.
I have also learnt that with all people-related matters, even when the best of tools are deployed, there are shades of grey where situational context and managerial judgment is required. Once we almost took a decision to eliminate a sourcing partner because all the hired people data (on performance, attrition) pointed to a bad experience. On closer examination and after holding talks with them, we learnt of other variables that we had not previously considered, such as the manager the underperforming person was assigned to, the territories that had been allocated and so on. This put things in a totally different perspective.
Seeing the larger picture, connecting separate data streams, doing deeper analysis, and understanding data context involve, no doubt, much more work than just pushing buttons and getting answers on a computational system. As with all things powerful, the real play is always in the intersection of smart technology and managerial effort.
Disclaimer: This is a contributed post. The statements, opinions and data contained are solely those of the individual authors and contributors and not of People Matters and the editor(s).