How to analyze distribution performance

How to analyze distribution performance: compare on-time delivery, unit fill rate, inventory turnover and distribution cost per completed order by site for the same period. Rebuild each measure from exported counts and amounts, then read the measures together. A low-cost site with incomplete deliveries needs a different response from a reliable site carrying too much stock.

1. Define the scorecard before comparing sites

Use a common calendar period, customer delivery window and cost boundary. This example measures physical distribution at a fictional US industrial supplies distributor for September 2026.

On-time delivery = Orders first received within the agreed delivery window / Orders due

Unit fill rate = Requested units supplied on the first dispatch / Requested units

Monthly inventory turns = Monthly cost of goods sold / Average inventory at cost

Distribution cost per order = Defined distribution expense / Completed parent orders

Here, on-time delivery is a date-only test: a partial first receipt can pass. Unit fill measures quantity on the first dispatch for those same due orders, through September 30. Cap supplied units at each line's requested quantity so an overshipment cannot conceal another line's shortage. Keep overdue orders without receipts in the due-order denominator.

Use the original agreed window consistently. The separate guide to OTIF, fill rate and on-time delivery explains the alternatives; this scorecard keeps date and quantity separate.

Completed orders belong to the month's completion cohort, which can differ from the due-order cohort. Count each parent order once when its final shipment leaves. Include fulfillment labor, packaging, outbound freight and assigned warehouse overhead; exclude merchandise cost and inbound freight already capitalized into inventory. Match expenses and accruals to completed orders under a documented policy.

2. Bring the exported summaries into two sheets

Start with the order register, dispatch and customer receipt summaries, inventory valuation reports and distribution expense export. Retain order IDs, site, requested dates, receipt dates, quantities and expense references in the underlying files. Resolve duplicate shipment rows and unmatched sites before aggregating.

Put the following service summary in a sheet named Service, with headers in row 1 and sites in rows 2 through 4. Columns A:F follow the table order. Counts are as of September 30, 2026; orders canceled before entering the due cohort are excluded consistently.

If an export has several rows per site, aggregate the counts before dividing. For example, an export on RawService with site in A, period-start date in B and due-order count in C can populate Service!B2 with:

=SUMIFS(RawService!C2:C1000,RawService!A2:A1000,A2,RawService!B2:B1000,DATE(2026,9,1))

This assumes nonoverlapping summary rows, not an order count repeated on every shipment. Google's SUMIFS documentation explains summing values that satisfy multiple criteria. Expand the ranges to include every exported row.

Put the resource summary in Resources, again using rows 2 through 4 and columns A:E in the order shown.

Use inventory at cost throughout. Purchased-merchandise cost includes acquisition charges under the cost method described in IRS Publication 538; substituting retail value changes the comparison. For material fluctuations within a month, use consistently spaced inventory snapshots rather than only the opening and closing balances used here.

3. Calculate the site results

Create a Scorecard sheet with the same site order. These Google Sheets formulas populate columns B:F in row 2; fill down through row 4. A zero denominator returns an unavailable result for investigation rather than a misleading zero.

B2: =IF(Service!B2=0,NA(),Service!C2/Service!B2)
C2: =IF(Service!D2=0,NA(),Service!E2/Service!D2)
D2: =(Resources!B2+Resources!C2)/2
E2: =IF(D2=0,NA(),Resources!D2/D2)
F2: =IF(Service!F2=0,NA(),Resources!E2/Service!F2)

Format B:C as percentages. Calculate with full precision; display percentages to one decimal, turns to two decimals and cost per order to cents. Covirage is not affiliated with Google.

Cleveland's steps are 940 / 1,000 = 94.0% on time; 19,200 / 20,000 = 96.0% fill; and (180,000 + 220,000) / 2 = $200,000 average inventory. Its monthly turns are 400,000 / 200,000 = 2.00, and cost per order is 24,500 / 980 = $25.00.

These are monthly turns. The inventory turnover ratio guide covers period selection and valuation; do not compare this column directly with annual turns.

4. Prove the network totals

Calculate network rates from summed inputs. In Scorecard row 5, use:

B5: =SUM(Service!C2:C4)/SUM(Service!B2:B4)
C5: =SUM(Service!E2:E4)/SUM(Service!D2:D4)
D5: =SUM(D2:D4)
E5: =SUM(Resources!D2:D4)/D5
F5: =SUM(Resources!E2:E4)/SUM(Service!F2:F4)

The network results are 1,888 / 2,000 = 94.4% on time, 38,440 / 40,000 = 96.1% fill, 880,000 / 560,000 = 1.57 monthly turns and 55,100 / 2,000 = $27.55 per order. Averaging the site on-time percentages gives 94.0%, which weights the smallest site like the largest.

Reconcile the 2,000 due orders and 2,000 completed orders to their separate registers; equal totals do not prove identical orders. There are 2,000 - 1,888 = 112 orders without a qualifying on-time receipt, and 40,000 - 38,440 = 1,560 units unfilled on first dispatch. Tie opening and closing stock to valuation reports and COGS and distribution expense to the included ledger accounts.

Finally, the cost identity must hold: 980 × 25 + 600 × 30 + 420 × 30 = $55,100. The cost per order guide covers the expense boundary and split-shipment checks.

5. Turn the comparison into an investigation

Reno has the strongest unit fill but the weakest delivery timing. The transport lead should inspect customer receipt exceptions, carrier lanes and dispatch timing before buying more stock. These totals identify the question, not its cause.

Dallas combines the lowest fill with the largest average inventory and lowest monthly turns. The inventory planner should match shortages to product-level stock and inventory aging by site. Excess stock in unrelated products will not fill the missing lines.

Cleveland has the lowest cost per order. Finance and the site manager should compare order size, handling and freight distance before treating its rate as a target for the other sites. Review agreed actions against the next comparable period, using customer commitments rather than an unsupported universal benchmark.

Where it goes wrong

• Dispatch called delivery. A warehouse departure timestamp does not establish customer arrival within the agreed window.
• Changing cohorts. Dropping unshipped due orders improves service artificially; counting shipment rows as completed orders understates cost.
• Mixed units or boundaries. Cases and individual units need consistent conversion. Inbound inventory cost and outbound distribution expense need separate treatment.
• Averages of rates. Network service, turns and cost require their own weighted denominators, not a simple average of site results.
• Targets without context. Product availability, handling requirements and order mix can explain differences that the summary alone cannot resolve.

Analyze your own distribution file in Covirage

Bring a sample of your own warehouse, delivery, inventory and expense summaries; Covirage maps the columns, and its tools compute and check the figures. The external AI model explains the result with every figure traced to its rows. Create a free Covirage account and bring a sample of your own data.

See Covirage for supply-chain teams for the operating context.