PAR Level: Definition, Formula, and How to Calculate It

PAR level (Periodic Automatic Replenishment level) is the minimum quantity of an item that should be on hand at the start of an order cycle to meet expected demand without stockouts. It is the target inventory level that a replenishment system orders — you reorder enough to bring inventory back to PAR.

The formula

The conceptually correct PAR level has three components:

PAR = base demand + statistical safety stock + manual buffer

where:

• Base demand = expected consumption between orders = consumption rate × order frequency days
• Statistical safety stock = z × σ × √(order frequency days), where z is the z-score for your target service level (0.67 for 75%, 1.28 for 90%, 1.65 for 95%) and σ is the standard deviation of daily demand
• Manual buffer = an optional fixed amount you add for items where stockouts are particularly costly (a key ingredient, a flagship product)

Worked example

A coffee shop sells 18 lbs of beans per day on average, orders weekly (7-day cycle), with a daily-demand standard deviation of 4 lbs and a 90% service-level target.

• Base demand = 18 × 7 = 126 lbs
• Statistical safety stock = 1.28 × 4 × √7 ≈ 13.5 lbs
• PAR ≈ 140 lbs

If you currently have 32 lbs on hand, the order recommendation is 140 − 32 = 108 lbs, rounded up to the next pack size.

PAR with decay

The simple formula above assumes no spoilage. For perishables — produce, dairy, fresh proteins — PAR has to account for the fraction of inventory that becomes unusable each day even if not sold. Decay rate models this.

With decay rate d, the base demand integrates over the order cycle as:

baseDemand = (s/d) × (s^(−T) − 1) × c
  where s = 1 − d, T = order frequency days, c = consumption rate

This is the exact formula LineNow uses. It correctly captures that perishables need to be ordered more aggressively because some inventory will not survive to be consumed. For a 7-day cycle with a 5% daily decay rate, the effective PAR is about 17% higher than the no-decay baseline.

What PAR is not

PAR is often confused with related concepts. To be precise:

• Reorder point is the inventory level at which you place an order. It is below PAR. See reorder point.
• Safety stock is just the buffer above expected demand. It is one component of PAR, not the whole.
• Maximum stock level is sometimes used interchangeably with PAR but typically refers to the upper bound of inventory that triggers price markdowns or returns; in the food-service literature, PAR is the target, not the cap.
• Min/max is a related heuristic where "max" is similar to PAR. PAR with statistical safety stock is more rigorous than min/max.

Why most PAR levels are wrong

In the artisanal procurement stack, PAR is set by the operator's gut feel, written on a whiteboard, and updated semi-annually when something goes wrong. This produces two failure modes:

1. Over-ordering on slow movers, because the operator buffers against rare events with a thick safety stock.
2. Stockouts on fast movers, because consumption rates change faster than the whiteboard.

The fix is not to tune PARs more often. It is to derive them from data. With real-time POS sync providing daily consumption, and a statistical formula that incorporates demand volatility, PAR levels can be recomputed every night automatically and stay current without manual intervention.

How LineNow computes PAR

For every line item, every day, LineNow:

1. Pulls the 30-day daily-bucketed sales sequence from your POS.
2. Computes mean (μ) and standard deviation (σ) of daily demand.
3. Classifies the demand pattern (smooth, intermittent, erratic, lumpy) using CV² and average demand interval.
4. For non-smooth patterns, applies the Syntetos–Boylan Approximation forecast.
5. Computes statistical safety stock with a z-score derived from your configured rush sensitivity (low = z 0.67, medium = z 1.28, high = z 1.65).
6. Integrates over the order cycle with explicit decay handling.
7. Adds any manual safety buffer you've set.
8. Surfaces a single PAR number, with the eight-line statistical justification available on click.

You can override any input. The system will recompute. The default is right more often than gut feel because it has more data.