The short version
- Track a small set of clear measures, understand how they’re calculated, and use them to improve the work.
- Use your own order and inventory data when comparing options.
- Agree the details with your provider before work begins.
Why the definition matters more than the number
Before comparing results, agree on the start time, finish time, exclusions, and data source for each measure. Two similar labels can describe very different calculations.
Order accuracy rate
Order accuracy measures whether the customer received the correct products and quantities. Track the reason for each error so the team can address the cause.
On-time ship rate, what counts as shipped
Agree when an order is due to ship and what counts as shipped. Separate warehouse processing from carrier transit and record any agreed exclusions.
Order processing and receiving time
Track how long orders take from release to shipment and how long deliveries take to become available inventory. Agree when each clock starts and stops.
Inventory accuracy, and the count process
Compare system records with physical counts and investigate differences. Nile uses pick-line cycle counts and a rolling quarterly physical audit to support accurate stock records.
Perfect order rate how the measures fit together
A perfect order arrives with the right items, in the right condition, with the required paperwork and timing. It brings several service measures together.
Scroll the table sideways to see every column
| Each component | Perfect order rate | Failures per 10,000 orders | Practical read |
|---|---|---|---|
| 99% | 96.06% | 394 | Four strong numbers, one mediocre outcome |
| 98% | 92.24% | 776 | About one order in thirteen has a problem |
| 97% | 88.53% | 1,147 | About one order in nine |
| 95% | 81.45% | 1,855 | Nearly one in five, and a support team on fire |
Cost per order, split into handling and carrier
Show warehouse handling and carrier spend separately, then look at the combined total. Keep the same cost categories in each reporting period.
- Units per order, which drives pick labor and carton size at the same time
- Average billable weight versus actual weight, which is your dimensional weight exposure
- Zone mix, which moves when you add or remove a fulfillment node
- Accessorial spend as a percentage of base freight, which is where surcharges hide
- Storage as a percentage of total fulfillment cost, which is your inventory discipline showing up as a bill
Return rate and return-to-sellable time
Track why products return and how long approved stock takes to become available again. This helps separate product issues from packing or shipping issues.
Retail delivery performance and customer delivery questions
On-time, in-full (OTIF) delivery rates and customer questions tell you different things. Use the retailer’s current definitions and review customer feedback alongside shipment records.
The full scoreboard
Build a short report your team will actually use. Give every measure a definition, a data source, a review frequency, and a person responsible for follow-up.
Scroll the table sideways to see every column
| Metric | Formula | Most common gaming method |
|---|---|---|
| Order accuracy rate | Orders with zero defects ÷ orders shipped | Counting lines instead of orders |
| On-time ship rate | Orders shipped by promised date ÷ orders due | Stopping the clock at label creation |
| Order cycle time | Order receipt to first carrier scan | Leaving out held orders and orders with problems |
| Receiving time | Arrival to inventory available for sale | Starting the clock at appointment time |
| Inventory accuracy, units | 1 − (sum of absolute variance ÷ units counted) | Netting overages against shortages |
| Inventory accuracy, locations | Locations with zero variance ÷ locations counted | Counting only recently replenished bins |
| Perfect order rate | Orders on time, complete, undamaged and documented ÷ orders | Dropping the weakest component |
| Cost per order | Total fulfillment cost ÷ orders shipped | Quoting pick and pack only |
| Return rate | Returns ÷ orders shipped from the same group of orders | Comparing returns with an unrelated group of orders |
| Return-to-sellable | Dock receipt to available-to-allocate | Stopping at box open rather than disposition |
| OTIF | POs delivered on time and in full ÷ total POs | Using your ship data instead of the retailer’s scorecard |
| Delivery-question rate | Order status contacts ÷ orders shipped | Counting only questions tagged as delivery inquiries |
Agreeing on service standards in writing
Agree service measures and review procedures with your provider. When a result misses the target, record the cause, the action, and when the team will check progress.
Scroll the table sideways to see every column
| Clause | What to write | What happens if you skip it |
|---|---|---|
| Definition | The exact start and stop events, the system of record for each timestamp, and the unit of measure: orders, lines, units, or cases | Both sides report different numbers from the same month and neither is lying |
| Measurement window | Calendar month, business days defined, holiday treatment, and whether peak weeks are measured separately or against a different target | Performance averages a quiet January against a broken November and November never shows up |
| Exclusions | A closed list, each capped as a percentage of volume, with excluded orders reported by count and reason | A 99% on-time rate that silently excludes the 6% of orders that ran late |
| Reporting schedule | Weekly order-level export plus a monthly summary, delivered by a fixed date, with a defined dispute window | You learn about a bad month six weeks later, after your customers already knew |
| Remedy | Proportional credits applied automatically, re-ship at the provider’s cost including expedited freight, escalation at a defined threshold, a cure period, and a termination right if the cure fails | The KPI is a target rather than an obligation |