A 3PL SLA scorecard that measures what your customers actually experience

A useful 3PL service level agreement needs measurable definitions for on-time shipping, order accuracy, inventory variance and receiving. Each measure should identify the orders or stock included, the clock being used, the evidence required and the person responsible when something goes wrong.

Start with five measures and the records behind them. A percentage without a denominator can conceal late orders, excluded shipments or a backlog that never appears in the report.

The definitions below are a practical starting framework to agree with a fulfillment partner. The example results are illustrative, not industry benchmarks or advertised 247 Fulfillment service levels.

Define eligible orders before measuring on-time shipping

An eligible order is an order that meets the agreed conditions for its promised shipping deadline. Define those conditions explicitly: released to the warehouse, payment or fraud holds cleared, required information complete, and inventory available under the agreed rules.

Record the eligibility timestamp, promised ship deadline, warehouse time zone and applicable working calendar. Orders released after cutoff should receive the next applicable deadline. The shipping cutoff planning guide explains how to connect that promise to actual operating capacity.

For an order-based weekly measure, use:

On-time shipping rate = eligible orders fully handed to the carrier by their deadline ÷ all eligible orders whose shipping deadline falls in the week × 100.

Count an order as on time only when all parcels required for that promise meet the deadline. If you measure shipments separately, label that metric clearly and retain the order-level view.

The denominator includes eligible orders that are still waiting in the warehouse. Excluding unshipped orders is an easy way to make a poor week look better.

Decide what proves carrier handoff

A printed label doesn't prove that a parcel left the building. UPS explains that “Label Created” means shipment details have been received, while its later shipped status reflects possession and movement in the network.

Agree on acceptable evidence: a carrier acceptance event, or a documented collection record linked to the parcels when scans are delayed. Define how missing evidence is investigated and how a result is corrected later.

Keep warehouse-ready time, actual handoff and customer delivery as separate timestamps. If the warehouse was ready but the scheduled carrier failed to collect, the customer-facing miss remains visible. Assigning the cause determines who acts and how the contract treats the event; it shouldn't erase what happened.

Measure order accuracy with a defined review window

For a shipped-order cohort with an agreed observation period, use:

Confirmed order accuracy = orders with no confirmed fulfillment error ÷ all orders in that cohort × 100.

Define an error as a wrong item, missing item, incorrect quantity or another agreed warehouse-controlled failure. Count an order with several errors once in the headline rate, and record each error type in the investigation log.

For example, nine distinct incorrect orders in a cohort of 1,500 give an accuracy rate of 1,491 ÷ 1,500 = 99.4%.

Name the observation period, its end date and unresolved claims. Recent orders haven't had the same opportunity to generate a complaint as older ones. Label early results provisional and restate the original cohort when a later confirmed error arrives.

This measures known errors; it isn't proof that every other customer received a perfect order. Audits and customer reports should feed the same record without counting the same incident twice.

Use absolute inventory variance

Net adjustments can hide mistakes. If one location has two units missing and another has two unexpected units, the net difference is zero even though both records are wrong.

Use a consistent SKU-location-lot unit of comparison where lot tracking applies:

Absolute unit variance rate = sum of absolute differences between physical and recorded units ÷ total recorded units in the counted records × 100.

Freeze or reconcile movements during the count. State which records were counted; a targeted count of problem locations doesn't represent the whole warehouse.

For example, 20 units of absolute difference across records containing 10,000 recorded units produce 0.2% absolute unit variance. Report unexpected stock in zero-balance locations too. If the entire counted set has zero recorded units, report the discrepancy in units and mark the percentage undefined.

Pair the rate with the number of mismatched records and the value of affected stock. The cycle counting guide provides a practical process for investigating those differences.

Start the receiving clock at an observable event

Define dock-to-stock time from an agreed arrival event to inventory being verified and available for allocation. Specify whether the clock uses elapsed hours or operating hours, and what happens outside receiving hours.

If compliant paperwork is required, record physical arrival and the time documents become complete. Show any paused time separately, with a reason and owner. A missing barcode shouldn't make an inbound shipment invisible.

For a deadline-based receiving measure, divide receipts made available by their deadline by all eligible receipts whose deadline falls in the reporting period. Include receipts still unfinished after that deadline. Define whether a receipt means a shipment, purchase order or another unit, and retain that choice.

Use the inbound shipment checklist to agree on the information required before goods arrive.

Bring a scorecard that leads to decisions

Here is an illustrative weekly review. Receiving uses a hypothetical agreed deadline of 16 operating hours; your actual target should reflect the work and contract.

In this shipping example, 20 orders missed their deadline but subsequently shipped; 40 remain open. The same 60 misses stay in the week's denominator. Carry open orders forward in backlog reporting without counting them as newly due orders next week.

Set contractual targets after defining the workload, exclusions and evidence. In each review, assign every material issue an owner, corrective action and due date. Check the previous actions before adding new ones.

If you are comparing fulfillment arrangements, contact 247 Fulfillment with the measures and customer promises your business needs. A sample report and a few actual order timelines will make that conversation more productive than a headline percentage alone.