Blog · Territory, capacity and quota planning · Supply chain
S&OP KPI metrics connect forecast error, customer service, inventory, and supply plan attainment. Work through a product-family scorecard, check the totals, and assign each exception to the person who can resolve it.
S&OP KPI metrics show whether demand, supply, service, and inventory are moving together under the agreed business plan. Compare them by product family, using the forecast and supply plan approved before the period began. A favorable total can conceal a shortage in one family and excess stock in another.
An S&OP review should end with decisions about demand assumptions, available supply, inventory exposure, and the resources needed to close gaps. This calls for a shared view: NIST's supply chain guidance identifies coordination across forecasting, demand and sales planning, operations planning, and inventory management as part of supply chain improvement.
Keep the scorecard small enough for people to explain the tradeoffs. Forecast performance tells demand planning where assumptions failed. Service tells sales and operations whether customers received the promised result. Inventory shows what the business is holding to support that result. Supply attainment shows whether the agreed replenishment or production quantity arrived.
The separate S&OP forecasting guide covers how the demand plan is formed. Here the question is whether that plan worked and what the next review should change.
| Measure | Calculation for the review | Source rows | Accountable owner |
|---|---|---|---|
| Weighted absolute percentage error, or WAPE | Sum of absolute forecast errors / actual demand | Frozen item forecast and matched actual demand | Demand planning |
| Forecast bias | Sum of forecast minus actual / actual demand | Same forecast and actual rows | Demand planning with sales |
| Customer OTIF | Orders meeting both conditions / orders due | Due-date cohort and delivery results | Customer service and operations |
| Historical inventory cover | Closing usable stock / average daily demand | Stock snapshot and demand history | Inventory planning |
| Supply plan attainment | Good supply received / frozen supply plan | Approved supply plan and accepted receipts | Operations or procurement |
These are explicit definitions for this scorecard. Positive bias means overforecasting here. Another business may use the opposite sign, so write the convention next to the measure. For delivery rules, including which promised date counts, use the definitions in KPIs for supply chain teams.
Match forecast rows by item, site, demand month, and forecast version. Assess the forecast that existed before the outcome was known. The authors of Forecasting: Principles and Practice explain why accuracy must be assessed using genuine forecasts against observations unavailable when the forecasts were produced.
Keep accepted receipts separate from quantities merely scheduled or shipped by a supplier. For inventory, distinguish usable stock from damaged, blocked, or consigned quantities according to the business's definition. A quantity-based operating view also needs a separate valuation check: the Census M3 definitions distinguish month-end inventory value and its materials, work-in-process, and finished-goods stages.
A US industrial distributor reviews September 2026, a 30-day month. This compact example has one item-site combination in each family, with all quantities expressed in the same defined case-equivalent unit. Demand is fulfilled customer demand; both families' orders were delivered within the month, although some missed their promised dates.
The approved forecast and supply plan were frozen before September began. The inventory column is the usable closing snapshot. Orders due and OTIF orders are counts, separate from quantities.
| Family | Forecast cases | Actual demand cases | Planned supply cases | Good receipts cases | Closing stock cases | Orders due | OTIF orders |
|---|---|---|---|---|---|---|---|
| Filters | 12,000 | 10,000 | 11,000 | 10,400 | 4,000 | 200 | 190 |
| Valves | 8,000 | 10,000 | 10,000 | 9,000 | 6,000 | 100 | 88 |
| Total | 20,000 | 20,000 | 21,000 | 19,400 | 10,000 | 300 | 278 |
Compute errors before combining the rows. Filters has absolute error of |12,000 − 10,000| = 2,000 cases; Valves also has 2,000. The signed errors are +2,000 and −2,000.
| Family | WAPE | Bias | OTIF | Historical cover, days | Supply attainment |
|---|---|---|---|---|---|
| Filters | 20.0% | +20.0% | 95.0% | 12.0 | 94.5% |
| Valves | 20.0% | −20.0% | 88.0% | 18.0 | 90.0% |
| Combined | 20.0% | 0.0% | 92.7% | 15.0 | 92.4% |
For Filters, cover is 4,000 / (10,000 / 30) = 12.0 days. Supply attainment is 10,400 / 11,000 = 94.5%. For Valves, cover is 6,000 / (10,000 / 30) = 18.0 days, and attainment is 9,000 / 10,000 = 90.0%.
The combined calculations divide totals: WAPE is (2,000 + 2,000) / 20,000 = 20.0%; OTIF is 278 / 300 = 92.7%; attainment is 19,400 / 21,000 = 92.4%. Combined cover is 10,000 / (20,000 / 30) = 15.0 days. Percentages and cover are rounded to one decimal place after calculation.
The combined cover is meaningful here because the units share a defined basis. For unrelated products, keep cover at item or family level and review inventory value separately. A high cover figure for Valves does not prove the right items were available on the dates customers needed them.
The forecast total exactly matches actual demand, so combined bias is zero. Yet absolute error is 4,000 cases. That difference proves why offsetting signed errors cannot stand in for forecast accuracy. With more items, calculate absolute errors at the agreed item-site-period grain before rolling up to families.
Inventory also needs a movement check. There are no returns or adjustments in this example:
| Family | Opening stock cases | Good receipts cases | Shipment issues cases | Closing stock cases |
|---|---|---|---|---|
| Filters | 3,600 | 10,400 | 10,000 | 4,000 |
| Valves | 7,000 | 9,000 | 10,000 | 6,000 |
| Total | 10,600 | 19,400 | 20,000 | 10,000 |
Opening stock plus receipts minus issues equals closing stock in each row. The total stock reduction is 10,600 − 10,000 = 600 cases. The 300 due orders must also split into 278 OTIF orders and 22 misses; Filters contributes 10 misses and Valves contributes 12.
Reconcile those populations to the source exports. An internally consistent table can still exclude a warehouse or duplicate a receipt. Keep the cutoff and definition with the report using versioned metric definitions.
Filters combines overforecasting with supply below plan. Demand planning should explain the forecast assumption before operations is asked to recover the whole supply gap. Valves combines underforecasting, weaker service, and higher closing cover. Investigate item availability, receipt timing, and customer due dates before authorizing more inventory.
The supply shortfalls are 11,000 − 10,400 = 600 cases for Filters and 10,000 − 9,000 = 1,000 cases for Valves. Together they explain the 1,600-case gap against the supply plan. They do not, by themselves, explain every customer miss. Record the suspected cause, supporting rows, action owner, and next review date.
Bring a sample of your own frozen forecast, actual demand, supply plan, receipt, inventory, and delivery files to Covirage. Covirage maps the columns; its tools compute and check the comparisons, and the external AI model explains the result with every figure traced to its rows. Create a free Covirage account, set up your business, and bring those exports to the review.
See Covirage for supply chain teams for the business context.
Use the monthly planning cycle for the cross-functional scorecard, with a stated cutoff for actuals and the approved plan version. Review urgent service or supply exceptions more frequently. Keep those operating updates separate from changes to the plan used to assess performance.
Demand planning owns forecast measures, operations owns supply attainment, and customer service owns delivery results. Finance checks inventory values and financial consequences. The S&OP process owner brings those views together and records who will act on each exception before the next review.
Yes. Overforecasting one product can offset underforecasting another, leaving total signed bias close to zero. Keep absolute error and signed bias by product family and underlying item. The total alone cannot tell a planner which stock to buy or reduce.
Either can answer a useful question, but label the basis. Historical cover compares stock with recent consumption; forward cover uses future demand. Do not compare the two as though they were the same measure, particularly when the business has a strong seasonal pattern.