Driven by the rise of big data, HR Analytics is fast becoming “the next big thing” in some circles, particularly in workforce planning and recruiting. But around important areas of talent development, issues of data relevancy and accuracy should give pause to any broad-brush analysis of corporate HR data.

First, some perspective: At the heart of the big data revolution is the thorough and relatively fast analysis of large volumes of public data — mainly transactional and fact-based data (e.g., purchases, web site visits), and social data/commentary (e.g., analysis of keywords and phrases in searches and reviews). Analytics look at this transaction and social data, and deal with the issue of relevancy (i.e., does this data matter?). The question of data validity is not a major concern (i.e., is this transactional data accurate?).

But HR data is a different kettle of fish, so to speak — and the threat of making false assumptions is higher with different types of HR data.

For instance, HR data includes clear and factual information that can be analyzed. We know when people were hired and left the company. We know when they were promoted and moved to different positions. We also have skills data for completion of courses and qualifications. This is the easy win for HR analytics — workforce planning around filling positions with people with the basic requirements for a job has been the early focus for HR analytics. However, this is all focused around minimum skill requirements for a job — and addresses specific “hard” skills.

But creating a high performance organization is about having people excel, which is focused on how people do their job with the skills they have. Assessing that is generally addressed in organizations by performance reviews, 360 assessments, and informal feedback. Key here: How valid and relevant is the data that comes from those efforts?

The many flaws of standard performance management systems are well documented in books and articles. There are issues on both sides of the equation — do we know what employees capabilities really are? Do we know what capabilities are required to excel in a position?

Treating this data as fact and applying analytics to it will not give useful results. We need to fix the accuracy problem and get rid of the bad data if we are going to be able to use analytics for development.

The other possible future for talent development analytics is looking at actual online behavior rather than assessing it with instruments. There are already social analytical tools that look at email and social communications to analyze interactions, and how groups of people work together. That is the Brave New World of talent development — it will be interesting to see how far this is taken in the HR realm, and how the accuracy, fairness and legality of such methods can be addressed.