Each list item carries a memory allocation of 500 bytes per item. This is an estimate to allow for the item name, item code, parent and other built-in properties, along with internal indexing that is needed to give efficient look-ups. Custom properties are additional to this - the size for custom properties is calculated as if they were a 1-dimensional line item.
Generally speaking model size could be estimated with cell size only but if you want to be precise or have massive lists/properties you then need to consider them accordingly. This is probably why you might come across some exceptions
PS Do you need properties as opposed to system modules ? (re: best practices)
I tried same calc on 8 different models including this and it works on 7 . Close enough considering its estimate .
Out of these 7 one model is actually a dev copy of the same model on which it doesnt work ???? . i am using Alm and models are syncd . how can i get such huge cell count difference but model sizes are same ?
One of the reasons can be the number of list items as they don't add to the cell count but do occupy space (500 bytes per item). Same holds good for the number of items within each subset that you have across all lists-these also don't give you the cell count but eats up your space.
Absolutely, I meant to say that cells are not visible in the grid and size of the model can't be estimated based on the number of cells that an Anaplanner sees. As you rightly mentioned earlier, that size and cell count doesn't go hand in hand.