On this page
A par level is a standing decision about how much of something to hold. Like most standing decisions it was made under conditions that have since changed, and unlike most it is rarely reviewed, because nothing forces the review.
The symptoms are familiar. Running out of a product on a busy night. Cases of something that has not moved in months. Ordering that feels like guesswork despite everyone involved being experienced.
Why pars drift out of alignment
The mix changes gradually. Not dramatically, which is why nobody notices. A serve becomes slightly more popular over eight months and the par that supported it in January is short by September.
The pattern changes shape. A venue that used to sell evenly across the week develops a heavy Thursday. Weekly volume is unchanged, so nothing looks wrong, but the delivery schedule now leaves a gap in exactly the wrong place.
Pars were set from memory. Frequently the original number came from an experienced person's estimate rather than from data. That estimate was probably decent and it was still an estimate, and it inherited whatever the assumptions were at the time.
What a par should actually be built from
Three inputs, and most bars use one.
Consumption over a comparable period. Not last month, which mixes a quiet week with a bank holiday. The same days of the week over several weeks, so the number describes a Thursday rather than an average.
The delivery interval, including the worst case. A par has to cover the gap to the next reliable delivery rather than the scheduled one. If a supplier is late once a quarter, the par that assumes punctuality will produce a stockout once a quarter.
Variability, not just the average. A product selling six a night every night and a product averaging six with a range of two to fourteen need different pars. Setting both from the average guarantees the second one runs out regularly.
That third input is the one almost always missing, and it is the one that explains why a bar can be well stocked on paper and still run out.
A workable review
Quarterly, for the products that carry most of your volume. The long tail is not worth the effort and can stay wherever it is.
For each: what did we actually sell by day of week over the last twelve weeks, what is the realistic delivery gap, and how variable is it. Then set the par and write down the date, so the next review knows what it is comparing against.
That last part sounds trivial and is the difference between a review and a habit. Pars with no history attached get re-estimated from memory every time.
What this costs and where it goes wrong
The analysis is straightforward and the assembly is not. Getting twelve weeks of sales by product by day of week out of most point of sale systems is a genuine chore, which is why the quarterly review usually happens once.
The other common failure is setting pars to eliminate stockouts entirely. That is achievable and expensive: the cost is capital sitting on a shelf and an increased chance of waste on anything perishable. The right target is a stockout rate you have chosen, not zero.
CoreTAP reads the point of sale you already run, so the consumption pattern by day and by product is available without the assembly step. That does not set your pars for you, and it removes the reason the review never happens.