Wednesday, March 11, 2015

4 Things You MUST Remember When Launching a ‘People Analytics’ Initiative

1. Focus on HR data integrity (first) until the vast majority of people analytics consumers explicitly express confidence in the data behind the planned analytics. There could well be a history of cynicism in your organization around HR and Talent Management initiatives – PARTICULARLY those that relied on data integrity perceived as suspect (best case) or totally unreliable (worst case). Achieving and sustaining high HR data integrity requires a multi-faceted program unto itself, one which is arguably as important as launching any analytics dashboards. You can employ multiple data scientists, statisticians and the best analyzing software, but once the data is perceived as faulty, you are probably pouring money down the drain with even the most impressive analytics. [Contact me if you want to discuss what an effective HR data integrity assurance program looks like: SBGConsultingLLC@Gmail.com]

2. Ensure that the people analytics team has the appropriate mindset and displays the appropriate behaviors through formal training. I am not referring to statistics training. While that is important, an air-tight regression analysis can be performed by a college intern. The training I am referring to is training which emphasizes the importance of “thinking and acting like a scientist.” This means not having many preconceived notions about what the data will likely tell you. If you do, and the notions are strong, the data relationships are probably so obvious it’s not worth the time to develop the predictive frameworks and syndicate them (e.g., employees in the last year of a vesting schedule have a much greater chance of leaving within 6-12 months than employees in the first year.) Scientists tend to first eliminate non-predictive relationships between commonly grouped data elements -- vs. narrowly (and in a “self-fulfilling prophesy” type of manner) focusing on proving their hypothesis.

3. Remember that a single metric rarely tells a story as it is largely out of context. This is the value of analytics, which by definition, highlights the predictable relationship between different items. As an example, seeing that turnover is trending up, or employee engagement is trending down, and reacting to such metrics in isolation, will likely be a pointless exercise that adversely impacts management and/or HR Department credibility. In contrast, as an example, look to identify which types of training in conjunction with where an employee sits in their compensation range or relative to market: (a) will likely drive voluntary turnover up, and (b) even trump the turnover-related predictive value of having a poor people manager as a supervisor/manager.

4. Remember that the ability to reliably predict what drives or impedes employee productivity … as measured by revenue per employee or other means more relevant to your organization … will likely have a significantly greater financial impact on your organization than reducing employee turnover or improving employee engagement, although they are all related. This is where it gets complicated, but also where people analytics can really “earn its keep” … e.g., determining with confidence when these 3 logically, intuitively linked data points will NOT be correlated or predictive of each other, and discovering what else is going on.

Steve Goldberg March 2015