Quick Summary: Best-in-class operations hit 99.5% or better on both on-time metrics the 2025 WERC DC Measures Report tracks: on-time shipments, meaning orders off the dock and in transit at the planned time, and on-time ready to ship, meaning orders packed, documented, and waiting for pickup at the planned time. The second number is the 3PL on-time shipping benchmark that matters most to you, because it measures the part of the job you control. Report both, calculate them from timestamps rather than from memory, and read them per client as well as building-wide.
A client emails to say your on-time rate slipped last month. You pull your own report and it says 99.2%, which is fine. They are looking at 94%. Nobody is lying. You are counting two different things, and until you agree on which one, the conversation goes nowhere useful.
That gap is the whole problem with on-time as a metric. It sounds like one number and it is at least three, depending on where you start the clock and where you stop it. This guide gives you the published benchmark figures, the two definitions the industry actually uses, a formula that holds up in a quarterly review, and the levers that move the number you are accountable for.
You benchmark a 3PL's on-time shipping rate by comparing it against the 2025 WERC DC Measures Report, which sets best-in-class at 99.5% or better for both on-time shipments and on-time ready to ship. Measure your own rate from system timestamps, against the ship-by date on the order rather than the delivery date, and calculate it separately for each client before you average it across the building.
The reason to use WERC rather than a carrier scorecard is scope. Carrier data measures transit. WERC measures the warehouse. Those are different jobs with different owners, and a 3PL that benchmarks itself against a delivery statistic has taken on responsibility for a truck it does not drive.
The 2025 WERC DC Measures Report lists twelve top distribution center metrics, and two of them are about being on time. They are not interchangeable.
On-time ready to ship is the percent of orders ready for shipment at the planned time, meaning packaging is complete, shipping documents are done, and the order is staged for pickup. Best-in-class is 99.5% or better.
On-time shipments is the percent of orders shipped at the planned time, meaning off the dock and in transit to the final destination. Best-in-class is also 99.5% or better.
The distance between those two definitions is the carrier's pickup window. An order can be packed, labeled, and staged by 2 p.m. and still miss the shipment clock because the driver arrived late, skipped the stop, or ran out of trailer space. Your crew did everything right and the shipment metric still records a miss.
That is why on-time ready to ship is the number to manage against internally. It isolates warehouse execution. On-time shipments is the number to report alongside it, because it tells you and your client where the failures actually live: in the building or on the dock apron.
Neither one is on-time delivery. Delivery adds days of carrier transit to a number that started as a measure of your pick and pack floor.
The published figures below carry their source, so they can go into a client deck without hedging.
| Metric | What it measures | Best-in-class | Source |
|---|---|---|---|
| On-time ready to ship | Orders packed, documented, staged at the planned time | 99.5% or better | 2025 WERC DC Measures Report |
| On-time shipments | Orders off the dock and in transit at the planned time | 99.5% or better | 2025 WERC DC Measures Report |
| Internal order cycle time | Order received to order shipped | Under 3.36 hours | 2025 WERC DC Measures Report |
| Total order cycle time | Order placed to customer receipt | Under 6 hours | 2025 WERC DC Measures Report |
| Order-picking accuracy | Picks correct before shipment | 99.68% or better | 2025 WERC DC Measures Report |
| Dock-to-stock cycle time | Arrival to put away and available | Under 3.5 hours | 2025 WERC DC Measures Report |
Best-in-class on-time sits at 99.5%, so a rate in the low nineties sits well outside the top tier rather than just under it. Internal order cycle time under 3.36 hours is the operational condition underneath the on-time number: an operation that turns orders in three hours has slack to absorb a bad morning, while one running at eleven hours is one forklift breakdown away from a miss.
For the carrier half of the equation, parcel performance is measured separately. During December 2025, UPS delivered on time 97.2% of the time, FedEx Express 95.3%, and the U.S. Postal Service 94.1%, according to ShipMatrix data reported by FreightWaves. Those are peak-season figures and they were up year over year across all three.
Read that carrier data with one caveat, and it is the caveat that proves the point of this whole guide. ShipMatrix noted that UPS moved all of its noon-or-earlier time-definite commitments to 3 p.m. and added a day to more lanes than FedEx did. Some of the improvement is a changed goalpost rather than a faster truck. If your client SLA is written against carrier delivery dates, your on-time rate moves when a carrier redefines its own commitments, and nothing in your building changed.
The formula is simple. The counting rules are where the argument happens, so settle them in writing before the first invoice.
On-time ready to ship = (orders staged by the ship-by time ÷ total orders due to ship) × 100
Four rules make that number defensible:
Use the ship-by date on the order, not the date the order arrived. An order that lands at 4:55 p.m. against a 5 p.m. cutoff is tomorrow's order. If your cutoff is not written into the client agreement, every late-afternoon drop becomes a dispute.
Measure from system timestamps, not from a daily tally. The timestamp when the pack station closes the order is the honest record. A count assembled the next morning rounds in your favor, and your client will eventually notice.
Count partial shipments as misses. If an order shipped without one line because inventory was short, it was not on time and in full. Counting it as on time hides the stockout that caused it.
Hold exceptions in a named category rather than deleting them. Weather closures, a client's own late inventory, and a carrier no-show are real and they are not your pick floor. Keep them visible as their own bucket. An on-time rate with exceptions silently removed is a number nobody can audit, which makes it worthless in exactly the meeting where you need it.
When your report and the client's report differ, the cause is almost always one of four things, and they are worth checking in this order.
Different clocks. You are measuring ready to ship. They are measuring delivery. The gap between the two is carrier transit, and it belongs in the conversation as its own line.
Different cutoffs. Your system applies a 3 p.m. cutoff. Their orders release at 4 p.m. and they expect same-day handling. Neither party is wrong; the agreement is just silent.
Different denominators. You exclude canceled and held orders. They count every order they submitted, including the ones their own credit hold stopped.
Different exception handling. You removed the two days a hurricane closed the building. They did not.
All four are definition problems, which means they get settled once in writing and then reported from a single set of records both parties can see.
Once the definition is settled, the number responds to a short list of changes.
Release orders in waves against the carrier cutoff, not in arrival order. Working a queue first-in-first-out feels fair and it puts the 5 p.m. UPS orders behind the ones that ship tomorrow. Scheduling work backward from each carrier's pickup time is the single largest structural gain available, and it costs nothing but sequencing. Our guide to wave picking and order release covers how to set the windows.
Fix receiving before you touch picking. An order cannot ship on time if the inventory is still on the dock. WERC puts best-in-class dock-to-stock under 3.5 hours, and a receiving backlog shows up as a picking problem days later. Dock scheduling is usually the cheapest place to find those hours.
Let scanning catch errors before the pack station, not after. A mis-pick discovered at pack-out consumes the time you needed for the cutoff. Order fulfillment accuracy averages 99.5% with a WMS against 92% without one, according to Aberdeen Group research cited by Supply Chain Dive, and every avoided error is a rework cycle that never eats into the shipping window. The picking accuracy metrics worth tracking in a WMS show where those errors concentrate.
There is a fourth lever that is not operational at all: renegotiate any SLA written against delivery dates. If you are accountable for a number that moves when a carrier changes its commit times, the fix is contractual. Evaluating carrier performance across a multi-carrier mix gives you the evidence to make that case.
A building-wide on-time rate of 99.1% can contain a client sitting at 96%, and that client experiences only their own number. They do not care about the average, and they are right not to.
Per-client reading also tells you something the building number cannot: whether a miss is an operational problem or an account problem. A client whose on-time rate is low because their orders arrive after cutoff, or because their inventory runs short, has a different fix than a client whose orders sit because your prime locations are full. One is a conversation, the other is a slotting project.
This is the same doubling that applies to every metric in a 3PL fulfillment KPI roster, and it works only when client separation is native to how the data is captured rather than a filter applied afterward. Running a multi-client warehouse means every number has a per-account version, and the per-account version is the one your client renewal turns on.