Days of inventory on hand (DOH) can refer to a historical accounting ratio or an operational estimate of how long current stock may last. Label which one you mean. They use different inputs and should not be substituted for each other.
For a buyer deciding what to order, current usable stock and expected consumption help estimate coverage. Supplier lead time, incoming shipments, commitments and uncertainty then determine whether a new purchase is needed.
The two formulas
| Measure | Formula | Question |
|---|---|---|
| Accounting days of inventory | Average inventory at cost / period COGS × days in period | How much inventory was held relative to cost of sales? |
| Simple operational days of cover | Usable quantity available for the modeled demand / expected daily usage | How long might this stock cover that usage? |
For annual accounting data, use the actual number of days in that reporting period. An annual ratio often uses 365; a quarterly ratio should use the corresponding quarter's days. Zero COGS or zero usage needs explicit handling rather than division by zero.
Accounting example
Assume fictional average inventory of $45,000 and annual COGS of $360,000 over 365 days:
Accounting DOH = $45,000 / $360,000 × 365 = 45.625 days
Inventory turns = $360,000 / $45,000 = 8
Accounting DOH = 365 / 8 = 45.625 days
Use inventory and COGS with consistent scope and valuation. Opening-plus-closing inventory divided by two is a simple average; seasonal changes may need more frequent snapshots.
The result does not say that every item physically spends 45.6 days on the shelf. Nor can you generally obtain portfolio accounting DOH by averaging two item-level day counts without the relevant weighting. See inventory turnover for the reporting distinctions.
Operational example
Assume a fictional retailer has 80 usable units available for the demand being modeled and expects constant usage of 12 units per day, with no incoming supply or other depletion:
Simple coverage = 80 / 12 ≈ 6.7 days
That is a forecast under stated assumptions, not a guaranteed stockout timestamp. A promotion, missing inventory movement, closure or changing demand can alter the result.
Physical on-hand, usable stock and uncommitted available stock are different concepts. State how reservations are treated. Do not subtract committed demand from stock and then count the same commitment again in the projected demand.
Comparing coverage with lead time
If a new order would take eight days to become usable, the example projects a gap before that new supply arrives. But an existing order arriving tomorrow could change the decision. Review open orders and their confirmed quantities, timing and destination before creating a replacement.
If lead time is four days, the same 6.7-day cover estimate exceeds mean lead time by about 2.7 days. That difference alone does not prove an adequate buffer: the demand and supply uncertainty still matter.
Under a simplified continuous-review policy with fixed lead time and independent daily demand, a normal-approximation safety-stock model can express a reorder trigger in units:
Mean daily demand = 12 units
Daily-demand standard deviation = 3 units
Fixed lead time = 4 days
z = 1.65 for an approximately 95% cycle-service target
Safety stock = 1.65 × 3 × √4 = 9.9 units
Reorder point = 12 × 4 + 9.9 = 57.9 units
Equivalent demand coverage = 57.9 / 12 = 4.825 days
The reorder point is applied to the relevant inventory position, including on-order stock and outstanding commitments under the policy. The equivalent 4.825-day figure is not a universal alarm on physical shelf stock. Reorder point and safety stock explain the assumptions and the distinction between cycle service and fill rate.
A more variable demand history can require a larger buffer for the same service objective. It does not require a higher z-score merely because its SBC class changes; z represents the selected service target within this model. Classification thresholds also do not prove that a forecast is reliable.
Losses and changing demand
A simple quantity divided by usage calculation assumes a stable depletion rate. If the model separately represents fractional loss, one possible recurrence is:
I(t + 1) = (1 − d) × I(t) − c
Here d is the per-period fractional loss assumption and c is use per period. This is the recurrence used in LineNow's planning calculation. Keep the units, event order and fractional-day conventions explicit. With zero modeled loss and positive constant usage, simple coverage reduces to on-hand divided by usage.
A loss assumption is not a shelf-life or food-safety determination. Avoid adding the same waste twice if observed consumption already includes it. With no observed usage, the more useful result may be “insufficient evidence” rather than a precise depletion date.
What to inspect before changing an order
Check current usable stock, the date of its last reliable count, unit conversions, demand assumptions, confirmed incoming quantities and any accepted supplier changes. A single stock total across locations may hide a local shortage.
Use LineNow's inventory replenishment workflow to review those inputs alongside the proposed purchase. A living purchase order preserves changes in expected supply and the actual partial receipts that follow.
Demonstrate one item with a late shipment and another with sparse demand. Confirm how the forecast handles each case and which person reviews the recommendation. A days-of-cover number is useful when its assumptions and next action remain visible.