Big data analytics cannot capture everything and can sometimes make unhelpful assumptions. Random correlations in data may appear significant but are actually just noise. Readily available metrics for analyzing performance may not be the most important variables. For example, a manager forcing people to meet short-term targets could be ruinous long-term. Additionally, the strongest predictor of future behavior is past performance as evaluated by those who know the person well, which data cannot replace. As a manager, one should use analytics while also recognizing its limitations to avoid being misled.