By Bob Sutor
Businesses are going social inside their firewalls. Whether buying a third party tool, or building their own, they’re sold on the idea that giving employees a way to share their knowledge will improve everything from morale to productivity.
The McKinsey Global Institute reported that these social tools can lead to a 25 percent increase in employee productivity. But what is the employer learning? How can it listen in and find ways to improve its processes and strategies?
All these social tools, including the return of the humble survey, create data that needs analysis. The volume of data might be more than the management staff can handle and make sense of by simply reading it.
A company’s internal social networking system is ripe for data mining. Employees understand that if the company is providing this business tool, the company can look at what is said. The information can help determine employee reaction or understanding of a company message.
As for surveys, asking the right questions – not to mention, the right number of questions – about an organizational change or idea can also give a sense of employee sentiment. The delivery doesn’t have to be complicated: an online form, or even an email address to which feedback should be sent.
So, whether through a social network or a questionnaire, it’s about creating an environment where employees are comfortable speaking honestly and creating a conversation with the employer.
Crunching employee data & influencing the influencers
Companies first need to know what they want to understand better. For example, how do employees use their internal online helpdesk? Is it a good experience? They can get some measure of “helpdesk” usage from analyzing their company social network, and then drill deeper. A heat map could then show the trending subtopics under this term, both positive and negative.
Our workforce software identifies these topics and presents them as rectangles in shades of green and red, with color indicating sentiment, and rectangle size indicating the number of employees talking about a topic. Overlaid on a map, it can also indicate where these topics are cropping up. Broken down further, the data by location, business area, even job category could show a company how different groups of employees, in different geographies, vary in their experience with the helpdesk. Does something about the helpdesk need tweaking? If so, where? Now a business can find the answer.
All this analysis comes down to text and the auxiliary information about employee roles and locations. From the surveys and social networks, to those email responses, businesses need to analyze keywords and phrases to develop context. And this analysis is still difficult – natural language processing can still have trouble teasing out the subtlety of irony. So, the question often becomes one of what kind of analysis does a business want, and how often? Is once a month fine or does it need to be close to real time?
Real-time analysis can be good for quick keyword filtering on how employees are responding to a recent announcement or change, answering questions, such as “did employees understand the message and, if not, what more should the company say?”
But as anyone who uses Twitter knows, you can get too far down into the weeds and miss the big picture. The more data to mine, the more time the analysis will take. This is where influencers come in.
Social networks are not flat. Some employees create more content, and have more influence on their peers. So, rather than asking “what is the sum total of all employees’ thoughts” on some corporate message, the company may first ask the opinion of several of these influencers. If company leadership can explain to this small group why something was or will be done, then a better expressed message can be given to employees, amplified by the influential employees they trust.
For example, IBM’s first blogging guidelines for employees in 2005 came about because corporate communications asked a number of early employee bloggers to draft the policy. It was incredibly empowering at the time. These influencers said “Wow. They let us do this instead of just having some anonymous policy makers set some rules!” And the powers that be said “this seems reasonable. Let’s go with it.”
IBM chose those influencers for this work because they had really self-identified themselves because of their early and independent blogging. Their excitement to participate rendered results that then spread a belief and confidence in all employees that the rules were reasonable. What that group came up with also laid the foundation for what are now IBM’s social business guidelines.
Data augments, not replaces, good management
With all of this analysis and heat mapping, the question of “will a computer make HR decisions?” always comes up. In short: No. Employees aren’t going to be whittled down to a score. Employees are important resources and talent. A machine won’t give anyone an excuse for poor communication of what employees need to understand to do their jobs. Analytics for HR does not replace good management nor an understanding of the social and cultural aspects of a working environment.
Analysis should improve upon and make sense of the available information, and how new information should become available and used. If there are things happening at a macro-level, such as online helpdesk improvements, or social media policy updates to keep up with the latest trends, the data will show this, and help a business decide what to do. Analysis bridges the gap between how employers make these decisions, and how employees respond to it. A field that has historically relied on psychology or “gut” instinct is becoming increasingly data-driven, turning hunches into substantiated assertions and helping make what was murky or unclear easier to see and process.