Three terms that get used interchangeably and mean different things. Getting them straight is most of the work.
The definitions
Reorder point. The stock level at which you place an order. It answers when. It is not a minimum you want to keep on hand; it is the level at which ordering has to start so you do not run out before delivery arrives.
Min / max. A pair. Order when stock hits min, order enough to reach max. Min answers when, max answers how much. Simple, visual, and the reason bin cards work.
Par level. A single target. Count what is there, order the difference, on a fixed schedule. Common in cribs and consignment because it suits a rounds-based replenishment visit rather than event-based ordering.
Reorder point and min are close cousins. Par is a different animal, because it is time-driven rather than level-driven.
Working out a reorder point
The base calculation is average usage during the lead time.
If you use four hose assemblies a week and the lead time is three weeks, twelve are consumed while you wait. A reorder point of twelve leaves you at exactly zero when the delivery lands, which is fine on paper and never fine in practice.
So you add safety stock, and the size of it depends on two things: how variable your usage is, and how variable the lead time is. Steady usage with a reliable supplier needs very little. Lumpy usage with a supplier who slips needs a lot.
The common error is padding safety stock to cover a lead time problem. If your supplier is unreliable, you are paying to hold inventory that fixes their process. Sometimes that is the right trade. Make it deliberately.
Setting max, and why it is usually wrong
Max is set by order economics, not by shelf space, and shelf space is what most people use.
It should reflect the cost of placing an order against the cost of holding stock. Cheap ordering and expensive holding means order little and often. Expensive ordering, whether through minimum order quantities, freight breaks or admin, means order more, less often.
Two constraints override the arithmetic. Pack quantity, because ordering seventeen of something sold in boxes of twenty-five is not a real option. And shelf life, which is where distributors of rubber goods and rated equipment get caught. If an item has a usable life from manufacture, a max that represents fourteen months of usage guarantees write-offs, whatever the order economics say.
When par levels suit better
Par works when somebody visits on a schedule and replenishes what is missing.
It is administratively simple, it suits VMI rounds, and it needs no usage history. What it cannot do is respond to a spike between visits. If usage doubles the week after a visit, par will not notice until the next one.
The usual compromise is par for the routine consumables and reorder points for anything where a stockout stops production.
Reviewing the numbers
Every set of levels is a snapshot of a demand pattern that has since changed. Levels set two years ago are describing a plant that no longer exists.
Review quarterly and look at three things. Stockouts, obviously. But also items that have not moved at all, which are candidates for removal from the stocked list rather than for a lower min. And lead times, because those drift quietly and nobody updates the reorder point when they do.
Where tagging helps
All of this depends on knowing consumption, and most operations only know it at count time.
Tagging the bin gives you the draw at the moment it happens, which turns average usage from an estimate into a measurement, and makes variability visible rather than assumed. Variability is the input that actually sets safety stock, and almost nobody measures it.
Tagging the item as well as the bin adds the other half: what is currently installed and when it comes due. At that point you are not reacting to a level at all. You are ordering against known upcoming demand, which is the only version of this that does not require safety stock to absorb your own ignorance.