Blog · Procurement and supply chain · Supply chain
Choose a supply chain analytical method by the decision, the available files and the assumptions the evidence can support. Work service, inventory and cost measures from the same operating rows, then choose the next comparison.
Analytical methods for supply chain management should follow the business question: describe a change, locate its concentration, investigate a relationship or compare a proposed action. Start with service, inventory and cost figures that reconcile to operating files. More elaborate analysis cannot repair mismatched dates, incomplete orders or an undefined denominator.
The general supply chain analytics overview explains the broader discipline. Method selection is narrower: what calculation will support this decision, and what assumptions must be true for its result to mean anything?
Write the decision before choosing the calculation. “Understand supplier performance” is too broad. “Decide which supplier needs a delivery review before the next sourcing meeting” identifies an action, an owner and the evidence needed.
| Business question | Method | Evidence required | Assumption to challenge |
|---|---|---|---|
| What changed by supplier or site? | Descriptive comparison | Comparable order, receipt and cost exports | The populations and period definitions match |
| Where is the problem concentrated? | Segmentation and ranking | Supplier, SKU, site and product-family identifiers | Segments are useful and consistently coded |
| Which factors move together? | Diagnostic comparison or scatter plot | Matched observations of the outcome and possible drivers | Association has an operational explanation |
| What changes under an approved assumption? | Scenario analysis | Baseline figures and explicit changed inputs | Other drivers can reasonably be held constant |
| Did the intervention improve performance? | Controlled evaluation | Before-and-after records and a credible comparison group | Other changes do not explain the difference |
The NIST introduction to exploratory data analysis emphasizes examining structure, outliers and underlying assumptions. That supports a practical sequence: inspect the records, compare groups and understand unusual observations before adopting a complicated explanation.
A fictional US industrial distributor reviews a completed delivery cohort over a 30-day period. Each order has one receipt, and every order in this example has arrived. On-time means received no later than the original agreed delivery date.
Bring an order export with supplier, order ID and promised date; a receipt export with order ID and actual date; an expedite-charge export; and daily inventory values with period cost of goods sold. Retain the original promised date when dates are revised, or service can improve merely because the target moved.
The table condenses those source records. Lead-time days are summed from order release to receipt for the same orders. Inventory is the average of daily closing balances, assigned to the supplier's SKU group using a consistent mapping.
| Supplier | Orders due | On-time orders | Total lead-time days | Expedite cost (USD) | Average inventory (USD) | Period COGS (USD) |
|---|---|---|---|---|---|---|
| A | 80 | 72 | 400 | 1,200 | 120,000 | 240,000 |
| B | 20 | 16 | 160 | 800 | 60,000 | 60,000 |
| Total | 100 | 88 | 560 | 2,000 | 180,000 | 300,000 |
The inventory amounts and COGS must cover the same assigned merchandise. If stock can come from several suppliers and its origin is unknown, compare inventory by SKU group instead of inventing supplier attribution.
On-time delivery rate = On-time orders / Orders due
Average lead time = Total completed-order lead-time days / Completed orders
Expedite cost per order = Expedite cost / Orders due
Inventory days = Average inventory / Period COGS × Days in period
| Supplier | On-time delivery | Average lead time (days) | Expedite cost per order (USD) | Inventory days |
|---|---|---|---|---|
| A | 90.0% | 5.0 | 15.00 | 15.0 |
| B | 80.0% | 8.0 | 40.00 | 30.0 |
| Total | 88.0% | 5.6 | 20.00 | 18.0 |
For supplier B, the calculations are 16 / 20 = 80.0%; 160 / 20 = 8.0 days; 800 / 20 = $40.00 per order; and 60,000 / 60,000 × 30 = 30.0 inventory days. Percentages and days are rounded to one decimal; per-order dollars to two.
The total delivery rate is 88 / 100 = 88.0%, not the simple supplier average of 85.0%. Total inventory days are 180,000 / 300,000 × 30 = 18.0. The separate inventory turnover method explains the broader stock-efficiency measure.
For a spreadsheet with the source table in columns A:G and supplier rows 2:3, these formulas rebuild the total measures:
=SUM(C2:C3)/SUM(B2:B3)
=SUM(D2:D3)/SUM(B2:B3)
=SUM(E2:E3)/SUM(B2:B3)
=SUM(F2:F3)/SUM(G2:G3)*30
The proof is the summed source population: supplier order counts must equal the eligible order register, and allocated expedite costs must equal the matching ledger charges. Missing supplier codes belong in an identified unassigned group until corrected.
Descriptive analysis flags supplier B: lower service, longer lead time and higher expedite cost per order. It does not establish that B caused the problem. B may supply harder products, receive larger orders or serve a different site.
Segment next by product family, destination and order-size band. Compare like orders within those groups. A scatter plot can then examine whether larger orders are associated with longer lead times. As NIST's scatter-plot guidance explains, an association does not prove causality.
A scenario can quantify a proposed service change without claiming it will happen. If B's on-time orders increased from 16 to 18 while all order counts stayed fixed, company on-time delivery would become (72 + 18) / 100 = 90.0%, a rise of 2.0 percentage points. Record the changed input and ask operations whether the intervention could produce it.
If the decision concerns funding an intervention, move to supply chain cost-benefit analysis. If it concerns inventory investment and payment timing, use the supply chain cash bridge. Those questions need different evidence from a supplier scorecard.
The supply chain analyst maintains definitions and reconciliation. Procurement and operations validate explanations; finance checks cost attribution. The decision owner records which assumption justified the action and when actual outcomes will be reviewed.
Bring order, receipt, inventory and cost exports with the identifiers needed for the comparison. Covirage maps the columns, and its tools compute and check the supplied comparisons; the external AI model explains the results with every figure traced to its rows. Create a free Covirage account, which takes about a minute, then set up your business and bring a sample of your own data.
See Covirage for supply chain teams for the supporting business context.
Start with descriptive comparisons on reconciled records: what changed, where and for which supplier or SKU. Segment the change before fitting relationships or testing scenarios. An aggregate result often hides a shift in product mix, order size or site coverage.
Delivery records can reveal associations between lateness and factors such as order size or lead time. They do not establish cause by themselves. Check the operational explanation, compare similar orders and, where practical, evaluate a controlled change before assigning responsibility.
There is no universal row count. The needed history depends on the question, variation, seasonality and how finely results are segmented. Report the observation count and missing records, and avoid treating a small supplier cohort as a stable performance estimate.