The purpose of this project is to investigate if application of online sensor measurements can improve the use of Evolutionary operation (EVOP) tools in livestock farms by reducing manual recording and by observing intermediate problem measures at an early stage. An example with reduction of tail-biting is used as a case. Traditionally, the frequency of the "tail-biting" in a pen is used to indicate if a treatment reduces the occurrence of tail-biting or not. Power calculations have shown that to detect a treatment difference of 10% vs 20%, 6 batches with 16 pens per section are needed to yield a power of 80%. It is generally accepted that there will be behavioural signs before an outbreak of tail-biting, such as increased activity (restlessness) and likely changes in the circadian rhythm. Using those indicators, based on online measurements of water consumption, it is our hypothesis that we can reduce the number of batches or pens needed and still get the same power. This is based on the assumption that early indicators are more sensitive than the event "tail-biting". To test this hypothesis, power calculations for EVOP design based on sensor measurements of diurnal variation in water use are presented. The calculations are made with typical farm dimensions and are based on estimates of standard deviations and variance components from a setup where water use was measured at pen level.