Line Fill Rate vs Case Fill Rate vs Order Fill Rate

The weekly operations review shows a fill rate of 93%. The key account manager's slide for the same week says 75%, and the customer's own scorecard says 40%. All three are correct. Each counts a different thing: cases, order lines or whole orders. The meeting ends in an argument about whose system is wrong, and nobody fixes the two SKUs that caused the shortages.

Once each rate is pinned to a stated unit, a stated denominator and a named source document, the three figures stop competing and start explaining each other. This guide runs case, line and order fill rate on one small dataset you can check by hand, so the gap between 92.9% and 40.0% becomes a diagnosis rather than a dispute.

• What the guide covers: the three formulas, a 12-line dataset with every intermediate number, value fill rate in US dollars and Indian rupees, denominator choices and edge cases.
• What you can reuse: an ownership table for the input fields, a SQL query that returns all three rates from one line table, and a checklist for publishing the definitions.

One shipment, three fill rates: why the numbers never agree

Case fill rate, line fill rate and order fill rate are not three versions of one number. They count shortages at three levels of granularity, so a single missing case weighs very little in the first, one whole line in the second, and one whole order in the third.

• Case fill rate: cases shipped divided by cases ordered, with shipped capped at ordered on each line. Some teams call it unit fill rate when the unit is the each.
• Line fill rate: order lines shipped in full divided by order lines ordered. A line short by one case counts as a failed line.
• Order fill rate: orders with every line shipped in full divided by orders received. One short line fails the whole order.
• Value fill rate: the value of capped shipped quantity divided by the value ordered, at a stated price list.

The ordering is nearly fixed. Case fill rate is usually the highest, order fill rate the lowest, and line fill rate sits between them. The distance between them tells you how shortages are spread: many small shortfalls across lines, or a few lines cut to zero.

The order-line dataset used for every calculation in this guide

All results below come from one day's outbound orders at an illustrative distribution centre: five orders, twelve lines, six SKUs, all ordered and shipped in cases. The US site prices in dollars and the Indian site prices the same SKUs in rupees from its own price list, not by currency conversion.

Case prices per SKU are A $18.00 or ₹1,250, B $42.00 or ₹4,600, C $9.50 or ₹720, D $26.00 or ₹2,100, E $64.00 or ₹5,300 and F $7.25 or ₹540. Three lines are short: SO-4102 line 1 by 8 cases, SO-4103 line 1 cut to zero, and SO-4105 line 2 by 2 cases.

Case fill rate: counting units shipped against units ordered

Case fill rate on this dataset is 209 ÷ 225 = 92.9%. It answers a supply question: what share of the volume customers asked for left the building. It is the right measure for inventory availability and labour planning, and the weakest one for customer experience.

The calculation looks trivial, but two rules decide whether it is trustworthy. Shipped quantity must be capped at ordered quantity on each line before summing, and every line must be in the same unit. Without the cap, ten extra cases of SKU D on one order would offset eight missing cases of SKU A on another and report a healthier number than the customers saw.

Case fill rate for a distribution centre equals capped cases shipped on the X12 856 ship notice or WMS pick confirmation divided by cases ordered on the X12 850 purchase order; on the twelve-line sample, 209 of 225 cases gives 92.9%, within 0.1 percentage point of a manual recount.

Unit of measure is the other trap. If SKU F is sold in cases of 12 but a customer orders 30 eaches, the line is 2.5 cases. GS1 assigns separate GTINs to each packaging level, so the order line and the pick confirmation can legitimately carry different identifiers for the same product. Convert both to one base unit, using the item master's pack hierarchy, before the line enters the calculation.

Line fill rate: when a line counts as filled and when it does not

Line fill rate on the dataset is 9 ÷ 12 = 75.0%. A line counts as filled only when capped shipped quantity equals ordered quantity, so SO-4105 line 2, shipped 48 of 50 cases, fails exactly as badly as SO-4103 line 1, shipped 0 of 6.

That binary rule is deliberate. A store replenishment buyer experiences a short line as a gap on the shelf or a follow-up order, regardless of how many cases arrived. Some teams soften it into "average line completeness", the mean of shipped ÷ ordered per line, which on this data is (9 + 0.80 + 0 + 0.96) ÷ 12 = 89.7%. That is a different measure and should carry a different name.

• Tolerance band: if contracts allow a short ship within, say, 2% of ordered quantity, record the tolerance on the line and publish the rate both with and without it.
• Split lines: when a WMS splits one order line into several pick tasks, roll them back to the customer's line number before counting, or one line becomes three.
• Zero-quantity lines: lines the customer cancelled before release leave both numerator and denominator; lines cut by the supplier stay in the denominator.

Order fill rate and why it falls fastest as orders get longer

Order fill rate on the dataset is 2 ÷ 5 = 40.0%: only SO-4101 and SO-4104 shipped every line in full. Three short lines, one in each of three different orders, failed three of five orders.

A quick model shows why order fill rate punishes long orders. If each line failed independently with the same probability, the chance an order is complete would be the line fill rate raised to the number of lines. With 75% line fill and an average of 12 ÷ 5 = 2.4 lines per order, that gives 0.75^2.4 ≈ 50%.

-- All three fill rates from one capped order-line table
WITH line AS (
  SELECT order_id,
         line_no,
         ordered_cases,
         LEAST(shipped_cases, ordered_cases)              AS shipped_capped,
         CASE WHEN shipped_cases >= ordered_cases
              THEN 1 ELSE 0 END                            AS line_complete
  FROM   order_line_fact
  WHERE  ship_window = DATE '2026-11-10'
    AND  line_status <> 'CUSTOMER_CANCELLED'   -- supplier cuts stay in
),
ord AS (
  SELECT order_id, MIN(line_complete) AS order_complete
  FROM   line
  GROUP  BY order_id
)
SELECT ROUND(100.0 * SUM(shipped_capped) / SUM(ordered_cases), 1) AS case_fill_pct,   -- 92.9
       ROUND(100.0 * AVG(line_complete), 1)                       AS line_fill_pct,   -- 75.0
       (SELECT ROUND(100.0 * AVG(order_complete), 1) FROM ord)    AS order_fill_pct   -- 40.0
FROM   line;

-- Independence check:  expected_order_fill = line_fill ^ (lines / orders)
--                      0.75 ^ (12 / 5) = 0.501

The actual 40% is below the model's 50% because the shortages were spread one per order rather than clustered. Had all three short lines sat in one order, order fill would have been 4 ÷ 5 = 80% with the same line fill. When order fill runs well below the independence estimate, the allocation logic is spreading scarce stock thinly; when it runs well above, shortages are concentrated in a few orders.

Value fill rate in dollars and rupees on the same twelve lines

Value fill rate weights each short case by its price, so it tracks revenue left on the table. On the dataset it is $4,204.00 ÷ $4,614.50 = 91.1% at the US price list and ₹3,34,370 ÷ ₹3,73,050 = 89.6% at the Indian price list, with identical quantities.

The gap comes from price mix. The six cases of SKU B cut on SO-4103 are 5.5% of ordered value in dollars ($252 ÷ $4,614.50) but 7.4% in rupees (₹27,600 ÷ ₹3,73,050), because SKU B is priced relatively higher on the Indian list. The same physical cut therefore costs more value fill at the Indian site. Neither market has a statutory definition of fill rate; the dual-market difference is entirely commercial.

Value fill rate in an ERP sales ledger weights capped shipped cases by the active price list; on identical twelve-line quantities, the US price list gives 91.1% and the Indian price list 89.6%, a 1.5-point spread caused by the six cut cases of SKU B carrying 5.5% of dollar value but 7.4% of rupee value.

Value fill rate also sits closest to finance. Short-shipped value is revenue deferred or lost, and in a 3PL it may also be handling revenue not earned. For how fulfilment activity turns into cost per order, see our guide to warehouse cost per order.

Denominators that change the answer: original, confirmed or revised

The quantity in the denominator must be named, because suppliers and customers routinely use different ones. The customer's purchase order, the supplier's acknowledgment and the customer's own later change can all carry different quantities for the same line.

• Original ordered quantity: as received on the X12 850 purchase order or keyed from a PDF or portal order. Most retailer scorecards use this.
• Confirmed quantity: as sent back on the X12 855 acknowledgment after allocation. Suppliers prefer it because it excludes cuts announced in advance.
• Revised quantity: after a customer-initiated change. Customer reductions should lower the denominator; supplier-initiated reductions should not.
• First-pass versus final: first-pass fill counts only the first shipment; final fill includes backorders shipped later within an agreed window.

The trace below is a composite from a consumer goods supplier, with identifiers changed. It shows how one denominator choice moved a monthly scorecard by nine points and nearly triggered a penalty discussion.

1. 03 Oct 2026, 06:10: retailer purchase order PO 7781204 arrives on an 850 for 1,000 cases across 40 lines.
2. 03 Oct 2026, 09:45: allocation cuts SKU 30155 by 93 cases; the 855 acknowledgment confirms 907 cases.
3. 05 Oct 2026, 14:20: the DC ships 880 cases, 27 short on picking; the 856 ship notice lists 880.
4. 04 Nov 2026, 08:00: the supplier dashboard reports case fill of 880 ÷ 907 = 97.0% against confirmed quantity.
5. 06 Nov 2026, 11:30: the retailer scorecard reports 880 ÷ 1,000 = 88.0% against the original order; both teams agree to publish both figures, labelled, from December.

Fill rate measured against the X12 855 confirmed quantity excludes supplier cuts announced before shipping; on purchase order 7781204, 880 shipped cases scored 97.0% against 907 confirmed but 88.0% against the 1,000 cases on the original 850, a nine-point gap from the denominator alone.

Edge cases: substitutions, over-ships, pack-size changes and cuts

Write the edge-case rules down before the first report goes out. Every one of them changes a numerator or denominator, and each is easier to settle in a definition document than in a monthly review meeting.

Choosing which fill rate to publish to each audience

Publish the rate that matches the decision each audience makes, and always show its definition next to the number. Operations steer on case fill, customer teams on line or order fill, and finance on value fill; nobody should see one of them unlabelled.

Our TMS evaluation built around one difficult transport order follows a single order's partial delivery through planning and billing. This guide steps back from one order to the population of orders, where the question is not how a partial shipment is executed but how often it happens and how it is counted. For wider KPI field definitions, see the supply chain KPI data dictionary.

Before go-live, run the checklist below against one week of real orders and reconcile the results by hand, as the twelve-line example does. If the hand count and the dashboard differ by more than 0.1 percentage point, fix the pipeline before anyone uses the number in a customer review.

The full library of logistics and supply chain guides covers the allocation, picking and carrier steps that move these numbers. If you are scoping a KPI pipeline that joins order, allocation and WMS data with agreed definitions, describe your order volumes and source systems in a request for proposal.