If an item is missing, its report can show few sales just when customers would have searched for it. Using that data to further reduce purchases is likely to confuse weak demand and unavailability of supply.
In store, unfulfilled requests can be noted in a simple way, distinguishing product, period and possible alternative. Online, actual availability must be compared with what the system shows. Visits to a page do not, however, coincide with lost orders.
Reordering requires a balance with overstocking. Increasing quantities indiscriminately absorbs money and can leave products unsold. I propose to examine delivery times, demand variability and the cost of unavailability for the most relevant categories.
A hypothetical example is an indispensable accessory to use a main product: its absence can affect more than one sale. Control then goes beyond the single reference. Recording periods without stock and requests received allows for better interpretation of data, without inventing a value of lost sales when sufficient elements are missing to estimate it.