A promotion can increase sales of a product and reduce those of a similar item. To evaluate it, one must observe the result of the category and the overall contribution, in addition to the success of the single reference. Local increase can hide an internal replacement.
I propose to compare quantity, margins and composition of the trolley in comparable periods. The advance on purchases also counts: a sale obtained today can be followed by a lower demand tomorrow. The observation window should cover this possible effect.
Let's think of a temporarily proposed package at a reduced price. Some customers might choose it instead of the usual packaging; others might add it to the cart. They are different behaviors, to be distinguished from the available data without attributing intentions that have not been observed.
The assessment remains cautious when changing assortment, availability or turnout. A useful promotion should have an explicit goal, such as trying a product or moving a stock. The final comparison verifies that goal and the cost incurred by the entire commercial offer.