If there is a gradual increase in cost on daly basis such increase will not be considered as an anomaly as the data would look like an expected increase however if we compare this cost over month to month the increase can be significant. Hence it is worth generating anomalies on such cost spikes on Monthly granularity.