Monitoring wind turbine (WT) performance offers a means of identifying abnormal operation, but only if natural disturbances of the operating regime change can be excluded. WT performance monitoring usually relies on the analysis of operational power curves, generally based on data from the supervision control and data acquisition system. However, these curves do not reflect the source of variability, negatively affecting the capabilities for detecting WT abnormal performance. This work aims at understanding and quantifying changes in WT performance variability due to different environmental conditions during normal and wake-free operating conditions, based on an offshore case study. The magnitude of performance fluctuations is highly influenced by environmental conditions, being higher during high turbulence intensity and low wind shear conditions. The Taylor law, with small time windows, is suitable to describe them for low-mid winds in the absence of dedicated wind measurements, often not permanently available offshore, and could potentially result in more effective performance monitoring solutions. Nevertheless, the heteroskedastic nature of the power deviations negatively affects fitting possibilities. The results support the importance of using low data aggregation periods to understand the dynamics of WT performance.