Every 3PL says its order accuracy is excellent, and most quote a number starting with 99. The trouble is that those numbers are measured in different ways, so two providers quoting the same figure can perform very differently. This guide explains what order accuracy and fulfilment error rates actually measure, what published benchmarks say, where picking and packing errors come from, how warehouses prevent them, and which questions to ask before you trust a provider’s figure.
Quick answer: For a well-run eCommerce fulfilment operation, expect at least 99.5% order accuracy measured per order, and ask for monthly figures that include peak season. The best distribution centres in WERC’s 2025 benchmark reach order-picking accuracy of 99.68% or higher. Before comparing providers, check three things: whether accuracy is counted per order, per order line or per unit; whether errors caught inside the warehouse and errors reported by customers are both counted; and over which period the figure is measured.
Jump to: What 1% means · Accuracy metrics compared · Why the denominator matters · Perfect order and OTIF · Benchmarks · Where errors come from · How warehouses prevent errors · What accuracy costs · Questions to ask a 3PL · When accuracy slips · FAQ
What one percent of errors really means
Percentages close to 100 hide the size of the problem. Turn the rate into orders and the difference between “very good” and “excellent” becomes clear. Each wrong order can mean a reshipment, a return, a refund, a customer service ticket and sometimes a bad review.
| Order accuracy | Wrong orders per month at 3,000 orders | Monthly cost at €12 per error (example) |
|---|---|---|
| 98.0% | 60 | €720 |
| 99.0% | 30 | €360 |
| 99.5% | 15 | €180 |
| 99.9% | 3 | €36 |
The €12 is an illustrative cost per error covering return or reshipment, replacement product and handling. Your own figure depends on product value, postage and how many customers you lose after a bad experience, which affects their lifetime value.
Accuracy metrics compared
“Accuracy” can mean at least five different KPIs. Use this table when a provider quotes a figure.
| Metric | Counts | Formula | Watch out for |
|---|---|---|---|
| Order accuracy | Orders shipped without any picking or packing error | Error-free orders ÷ all orders × 100 | Counting only errors customers complain about |
| Line accuracy | Order lines picked correctly | Correct lines ÷ all lines × 100 | Looks higher than order accuracy when orders have several lines |
| Unit (pick) accuracy | Individual units picked correctly | Correct units ÷ all units × 100 | Usually the highest number of the three |
| Perfect order rate | Orders on time, complete, undamaged and correctly documented | Perfect orders ÷ all orders, or the product of component rates | Loose definitions of “on time” or “damaged” |
| OTIF | Orders delivered on time and in full | On-time and complete orders ÷ all orders × 100 | Measuring despatch instead of delivery |
Inventory measures sit alongside these: fill rate shows how much of the ordered quantity could be shipped from stock, and inventory count accuracy shows whether the system matches the shelf.
Why the denominator changes everything
The same operation can produce three different accuracy figures depending on what you divide by. Take one month with 2,000 orders, 4,600 order lines and 6,800 units, and 14 errors, each on a different order:
| Measured per | Calculation | Accuracy |
|---|---|---|
| Order | 1,986 ÷ 2,000 | 99.30% |
| Order line | 4,586 ÷ 4,600 | 99.70% |
| Unit | 6,786 ÷ 6,800 | 99.79% |
Nothing changed in the warehouse, only the way of counting. For an online shop, per-order accuracy is closest to what customers experience: one wrong item in a four-item order still means an unhappy customer. If a provider reports only unit or line accuracy, ask for per-order figures or convert them using your average lines and units per order.
Perfect order rate and OTIF
Perfect order rate combines several measures. A common method multiplies the component rates, so a weakness anywhere pulls the total down. Four good-looking rates of 98.5% on time, 99.3% complete, 99.6% undamaged and 99.8% correct documents give a perfect order rate of about 97.2%: roughly three orders in every hundred fail somewhere.
Multiplying assumes failures are spread across different orders. When problems cluster in the same orders, the product understates real performance, so counting perfect orders directly is more precise if the data allows. Also check who is responsible for each part: the warehouse usually answers for picking, packing and documents, the carrier for transit time and damage in transit. A weak OTIF figure can be a carrier problem rather than a warehouse one.
Who counts an error, and when?
An error can be caught in three places: inside the warehouse (for example at the packing station), by the customer, or not at all. If a provider counts only customer complaints, its accuracy figure is really a complaints rate, because many customers who receive a wrong or missing item never report it, or simply return the order.
Benchmarks have their own definitions. WERC defines order-picking accuracy as accuracy of the picking process where errors can be caught and corrected before shipment, so internal catches are part of the picture. The most honest figure combines errors caught inside the warehouse with verified customer reports.
Benchmarks: what good looks like
The most widely cited source is the DC Measures study by WERC, the US Warehousing Education and Research Council, which surveys distribution centres every year and defines best in class as the top 20% of respondents. The full report is paid; the figures below come from a published summary of the 2025 report.
| Metric (WERC DC Measures 2025) | Best-in-class level |
|---|---|
| Order-picking accuracy | 99.68% or higher |
| On-time shipments | 99.5% or higher |
| On-time ready to ship | 99.5% or higher |
| Inventory count accuracy by location | 99.5% or higher |
| Dock-to-stock cycle time | Under 3.5 hours |
| Internal order cycle time | Under 3.36 hours |
Accuracy also depends on how picking is done. Supply chain consultancy MWPVL International gives typical gross error rates by picking technology:
| Picking method | Errors per 1,000 units picked | Pick accuracy |
|---|---|---|
| RF scanning | 5–7 | 99.3–99.5% |
| Pick-to-light | 3–5 | 99.5–99.7% |
| Voice picking | 0.3–3 | 99.7–99.97% |
These are per-unit pick figures from mostly larger distribution centres, not audited 3PL order accuracy. Taken together, a realistic expectation for a well-run eCommerce 3PL is 99.5% order accuracy or better, measured per order, with the best operations higher. Expect lower figures for ranges with many look-alike products, such as clothing sizes, cosmetic shades or spare parts, and for large baskets, because every extra item is another chance for an error.
Where fulfilment errors come from
Most people think of picking errors, but mistakes can start long before the picker and end after the parcel is packed.
| Stage | Typical errors | Main prevention |
|---|---|---|
| Goods-in | Wrong SKU or quantity booked in, unlabelled cartons, supplier mix-ups | Advance shipping notices, scanning and counting on receipt |
| Putaway and storage | Stock placed in the wrong location or mixed with a similar product | One SKU per location, scan to location, cycle counts |
| Picking | Wrong item, size or colour; wrong quantity; missed line | Scan every item, clear locations, separate look-alike products |
| Packing and labelling | Items or labels swapped between orders, wrong insert, wrong carrier service | Scan at pack, one order per bench at a time, weight checks |
| Data and integrations | Wrong barcode on a product, variant mapped to the wrong SKU, duplicate orders | Clean product data, checks after launches and integration changes |
Errors that start at goods-in or in your product data look exactly like picking errors to the customer, so a good root-cause analysis follows each error back to where it began. When you look at error causes, a few products or causes usually explain most of the errors, in line with the Pareto principle, so fixing those first gives the biggest gain.
How warehouses prevent errors
Barcode scanning at every step
Scanning is the basis of accurate fulfilment. When pickers scan the location and every item against the order in the warehouse management system, a wrong product is rejected before it leaves the shelf. It only works if every product has a unique, correct barcode, ideally a GTIN issued through GS1.
Verification at the packing station
A second scan when packing confirms that the right items go into the right parcel with the right label. A checkweigher that compares each parcel with its expected weight catches missing or extra units, which is useful when an order contains several of the same item.
Picking technology
Pick-to-light, voice picking and automated systems such as an automated storage and retrieval system guide people to the right place and quantity. They reduce errors further, but they only pay off at higher volumes.
Slotting and layout
Good slotting keeps look-alike products apart, for example two shades of the same lipstick in different aisles rather than neighbouring bins. Clear labels, good lighting and fast movers near packing reduce errors and fatigue. Layouts need review as product ranges change.
Error-proofing processes
The lean idea of poka-yoke designs mistakes out instead of relying on attention: the system won’t print a label until every item is scanned, or a predictable order such as a subscription box is pre-packed before orders arrive, leaving little to pick. Brands can help with distinct packaging for similar variants and written packing instructions for inserts and gift wrapping.
Training and peak staffing
Experienced, stable teams make fewer mistakes, and tracking errors by picker shows where coaching helps. Accuracy often drops in peak season when temporary staff arrive. An annual average can hide a weak November, so ask for monthly figures and how temporary staff are trained and supervised. Sharing forecasts early, for example before a flash sale, lets the warehouse plan staff instead of scrambling.
What accuracy costs
Each step closer to 100% usually costs more than the last. Moving from paper lists to scanning removes many errors at modest cost; checking every order twice by hand rarely pays.
Example: a scan at the packing station adds about 4 seconds per order. At a labour cost of €24 an hour that is about €0.027 per order, or €27 per 1,000 orders. If it cuts errors from 8 to 2 per 1,000 orders and each error costs €12, it saves €72 for every €27 spent. A full manual re-check of 25 seconds per order would cost about €167 per 1,000 orders to catch perhaps one more error worth €12, so it costs more than it saves.
Run the sums with your own cost per error. High-value, fragile or regulated products justify more checks than low-cost consumables.
Questions to ask a 3PL about accuracy
- What exactly do you measure: per order, per line or per unit? Can I see all three?
- Which errors count: internal catches, customer reports or both? Do you verify customer reports and add them to the count?
- Over what period: can I see monthly figures for the last 12 months, including peak season?
- How is each pick checked: is every item scanned, and is there a check at packing?
- How do you handle look-alike products and items without barcodes?
- What is in the SLA: which accuracy level is committed, how is it reported, and what happens if it is missed?
- How are errors attributed: warehouse, carrier or brand data, and who pays for each?
A provider that answers these clearly, in writing, usually has nothing to hide. Agree the definitions in your service level agreement, together with reporting frequency and any service credits.
What to do when accuracy slips
Start with evidence. Before raising the issue with your account manager, collect:
- The error rate by week or month, using the same definition each period.
- Individual cases with order number, date, SKU and type of error.
- Customer photos or messages confirming what arrived.
- Patterns by product, product type, weekday or shift.
- Changes on your side, such as new products, barcodes or integrations, that happened at the same time.
Then ask for a root-cause analysis and a written plan with dates. A wrong barcode can be fixed in days; staffing or layout problems can take weeks. Agree how improvement will be measured and when you will review it. If accuracy doesn’t recover in the agreed time, the SLA, service credits and exit terms come into play.
Switching provider carries its own risk, because a new team has to learn your products. If you do move, check accuracy weekly in the first weeks and consider moving stock in stages, running both warehouses in parallel for a short time so your shop stays live. For a wider comparison, see 3PL vs in-house fulfilment costs and boutique vs enterprise 3PLs.
The limits of accuracy figures
No warehouse reaches 100%. People pick products, data is never perfect and labels occasionally end up on the wrong parcel, so a provider claiming 100% is probably not counting everything. Aim for a low, stable and honestly measured error rate, set at the level where extra checks still cost less than the errors they prevent.
Accuracy figures are self-reported and hard to verify from outside. You can make them harder to flatter: agree definitions in writing, ask for monthly reports including peak months, run occasional test orders and compare the provider’s numbers with your own customer service data.
How Waredock states accuracy
Waredock is a 3PL, so apply the same questions to us. Our service pages state 99.9% order accuracy, and our standard SLA commits to at least 99.5% pick accuracy and 99%+ on-time despatch for orders received before the cut-off time. Ask us how each figure is measured and for the reporting that sits behind it, as you would with any provider.
Related: Picking · Pick and pack · Order processing · 3PL · Inventory counting 101 · Guide to warehouse planning · Inbound vs outbound logistics KPIs · Returns handling
Frequently asked questions
What is a good order accuracy rate for a 3PL?
For eCommerce fulfilment, expect at least 99.5% order accuracy measured per order. In WERC’s 2025 DC Measures benchmark, best-in-class distribution centres reach order-picking accuracy of 99.68% or higher.
How is order accuracy calculated?
Divide the number of orders shipped without any error by the total number of orders and multiply by 100. Line and unit accuracy use order lines or units instead of orders and give higher figures for the same performance.
What is the difference between order accuracy and perfect order rate?
Order accuracy covers picking and packing errors only. Perfect order rate also requires the order to arrive on time, undamaged and with correct documents, so it is always equal to or lower than order accuracy.
What causes most picking errors?
Look-alike products stored close together, unclear locations, picks that are not scanned, and wrong data such as incorrect barcodes or stock booked into the wrong location at goods-in.
Does order accuracy drop in peak season?
Often, because temporary staff are still learning the products and volumes are high. Ask for monthly accuracy figures that include peak months rather than an annual average.
Can I check my 3PL’s accuracy figures myself?
Partly. Compare the provider’s reports with your own customer service and returns data, run occasional test orders and agree the measurement definitions in your SLA.
Sources
Benchmarks are taken from the sources below and were checked in November 2026. Cost examples are illustrative.
