Restaurant Par Levels for Fresh and High-Variability Items: Why Static Pars Break Across Sites

Why One Fixed Par Level Breaks on Fresh and High-Variability Items

A six-location group's operations lead found their par levels worked fine for shelf-stable items but broke down for proteins and fresh grocery, where one site's weekly demand could swing by $10,000. A par level is simply the stock quantity that triggers a reorder, and it holds only while demand is steady. On fresh and high-variability items a number you set once is too high half the year and too low the other half.

The fix is not a looser number. It is to stop treating every item the same: keep static pars where demand is genuinely stable, where the standard par level formula still holds, and move fresh, high-variability items onto dynamic par levels built from recent sales. The rest of this guide is a decision tree for telling which item belongs on which branch, so the question for each item becomes simple: does its demand move enough to make a fixed number wrong?

Sort Every Item by Two Questions: How Perishable, How Variable

Every item in your store answers two questions that decide its par strategy: how fast does it spoil, and how much does its weekly demand move. Plot those two axes and four groups fall out, and only one of them is a problem.

Stable and shelf-stable items sit in the easy corner. Canned tomatoes, dry pasta and cleaning supplies barely move week to week and do not expire, so a static par you set once and check occasionally is exactly right. Steady perishables like daily bread and house milk expire fast but sell at a predictable rate, so they still take a static par, just a tight one on a short order cycle. Shelf-stable but variable items, such as bottled drinks tied to an event calendar, can ride a static par you review roughly monthly.

The problem corner is the last one: items that are both perishable and variable. Fresh proteins, seafood and leafy produce spoil in days and swing hard with covers, weather and season. As a rule of thumb, if an item's weekly demand moves more than about 20% between a quiet week and a busy one, a fixed number is wrong most weeks, and this is where dynamic par levels earn their keep. So the first question for each item is simply which of the four corners it sits in.

The Test: Is This Item on a Static Par or a Dynamic One?

The test is a single question you can run item by item: does this item's weekly demand swing more than about 20%, and does it spoil within a few days? If the answer to both is no, keep it on a static par and move on. If the answer is yes, take it off its fixed number and put it on a dynamic, forecast-driven par.

To answer honestly, pull the last eight weeks of usage for the item at one site and compare the busiest week with the quietest. If they are close, the item is stable. If they are far apart, it is variable, and a static par cannot serve both weeks. Picture a par of 40 units you set in the low season: it leaves you 50 units short the moment the same item peaks at 90 units in a busy week, so you either stock out or you carry dead stock the rest of the year to cover a peak that rarely comes.

Setting a Dynamic Par Without Guessing the Number

Once an item lands on the dynamic branch, the goal is to stop hand-guessing a new number and let recent demand set it for you. A dynamic par is rebuilt from what the item actually sold, not from a manager's memory of last season.

Supy's AI Sales Forecasting predicts daily sales by branch 14 days ahead, down to the menu item, and shows the forecast against an 8-week historical average, reaching under 10% variance between forecast and actual for a multi-site brand. That forecast feeds the par and minimum thresholds you hold per item per site, so the reorder point tracks real demand instead of a stale guess. From there, Fill to PAR auto-fills the order quantity from current stock against that par, and Predictive Ordering assembles a ready-to-submit purchase order that you still review line by line before it goes out. The number moves with the business, and no order ever leaves without a human check. Because thresholds live per item per site, you can manage all of this alongside the rest of your stock in one restaurant inventory management view.

Why the Same Item Needs a Different Par at Each Site

The last trap is copying one par across a group. The same fresh item can pull $9,800 a week at a busy airport outlet and $3,100 at a quiet suburban site, with two more branches somewhere in between. A single group-wide par is wrong at every site except the one nearest the average, so it drives over-ordering and waste at the small sites and stockouts at the big ones at the same time. Per-branch forecasting fixes this because each site gets its own model and its own par, rather than a headquarters number pushed down to all of them.

That gives you a clear place to start this week. Take your fastest-moving fresh category at your busiest site, pull the last eight weeks of usage, and if the swing is more than about 20%, take that item off its static number and set a demand-based par. If an item is stable and shelf-stable, leave its static par alone and spend the effort where the demand actually moves. You do not need to convert the whole store at once, only the items sitting in the perishable-and-variable corner where a fixed number was always going to be wrong.