I first became involved in energy monitoring and targeting when I worked in local government, at a time when it was impossible to acquire either meter readings or degree day figures more frequently than once a month. Monthly analysis and reporting was therefore the best we could achieve. However, when I became an independent consultant I acquired a number of energy-intensive industrial clients, notably in the paper and board industry, where weekly management cycles were the norm. Industrial users are much more likely to be disciplined about taking weekly meter readings, and just as importantly, they usually collect the relevant driving-factor data to match. The advantages of weekly reporting are twofold: (a) problems become evident more promptly, and (b) one is collecting four times as many data samples, which means that regression analyses can be carried out with a shorter time-span of historical data, and with greater confidence in the results.
Those same industrial users generally had some weather-dependent energy use, and that is what pushed me to develop weekly degree-day data collection. Week-by-week variation in degree day values is much more erratic than monthly variations but thanks to the way regression models work, that is not an issue, and abnormal behaviour around seasonal transitions can be seen much more clearly.
Some industrial process plants operate continuously and on such a scale that daily reporting and assessment is viable. I have even seen systems proposed which can report more frequently than that. “Real-time” exception reporting (with hourly or more frequent assessments) is a seductive idea. However, there is a downside to very frequent reporting, which is that transient effects and latencies begin to intrude and degrade the accuracy of your expected-consumption formulae so that you end up with just a barrage of spurious alerts.
My advice is generally to consider weekly reporting as the default starting point, but I wouldn’t be dogmatic about that. Monthly assessment could be appropriate and acceptable for monitoring buildings, while daily assessment can be a very good option for continuous process plants and energy-intensive buildings like data centres.