People analytics is tricky to do for many different reasons. There are:
- Technical challenges
The data that feeds people analytics tends to be predominately unstructured (text, audio, video, network), comes from dozens of disparate and semantically misaligned data sources, is notoriously ambiguous and prone to misunderstanding, is sparse with lots of missing data, requires complex analysis algorithms, and is BIG.
- Privacy challenges
Analyzing people data is a sensitive topic; few people like to be judged by others and are even less thrilled when its a machine doing the judging. Privacy ends up being the stick that people use to retaliate, however the issues are much more complex.
- Business challenges
Its absolutely critical that the appropriate business models are in place before analysis is generated and used by the business; methodologies are needed to ensure that the right analysis is applied to the right business problem.
Despite all these challenges, people analytics is doable and, if done sensitively, can provide significant value to both individual and organization. A few months ago I was asked to do a short interview sharing my own personal experiences of building a people analytics system within IBM that I wanted to share with you. If you are trying to build such systems within your own organization, there is light at the end of the tunnel and the results are most definitely worth the effort.