Wait time is the non-value time a part spends between operations, sitting in a queue because the next machine is busy, waiting for the rest of its batch to finish, waiting for a forklift, or waiting for inspection to release it. In most plants wait time is the largest single share of total lead time, which makes reducing it the fastest way to shorten delivery without touching the actual work.
The trap is that wait time is invisible on a work order. The order shows setup and run time; it does not show the four hours the tote sat between them. Because it is invisible on paper, wait time rarely gets a project until someone physically follows a part and clocks the gaps. This post names the four kinds of wait, explains why they dominate lead time, and gives a step order for cutting them, starting with the ones that cost nothing to fix.
What is wait time in manufacturing?
Wait time is any stretch where a part exists but nothing is being done to it. It is pure non-value-added time: the customer did not order the wait, and removing it changes nothing about the finished part except that it arrives sooner. If you have run a value-added versus non-value-added time tally, wait time is the bulk of the black band, the part of lead time that adds no value.
Wait time is different from cycle time which is the time a step actually takes to do its work. A five-minute operation can carry three hours of wait in front of it. Improve the five minutes and you save seconds; drain the three hours and you save an afternoon. That is why teams that measure only machine cycle time keep getting faster machines and no faster deliveries.
What are the four kinds of wait?
Naming the wait tells you which fix applies. Almost every delay on the floor is one of four types.
| Type of wait | What causes it | Primary lever |
|---|---|---|
| Queue wait | The next resource is busy; parts line up in front of it | Cut WIP, balance the line, cap releases |
| Batch (process) wait | A part waits for the rest of its lot before the lot moves | Shrink batch size, move toward one-piece flow |
| Transport wait | Parts wait for a mover, a truck, or a scheduled milk run | Shorten routes, move cells closer, raise move frequency |
| Inspection or decision wait | Parts wait for QA disposition, a hold release, or a signature | Inspect at the source, pre-authorize routine releases |
Queue wait and batch wait usually account for most of the total, and they are the two most people never measure. They feel like "just how the plant runs," which is exactly why they persist. Transport and inspection wait are more visible, you can see the tote sitting on the dock or the pallet in the hold cage, but they are usually the smaller share. The instinct to fix what you can see first is exactly backwards here: the biggest waits are the ones nobody clocks.
One caution before you attack any of them: not all wait is bad. A small, deliberate buffer of work in front of the constraint is protective, it keeps the plant's slowest, most important step from starving when an upstream machine hiccups. That buffer is business-necessary wait, not pure waste. The goal is to remove the accidental queues while keeping the one queue you put there on purpose.
Why does wait time dominate lead time?
Because of Little's Law: average lead time equals average work-in-process divided by throughput. If a line holds a lot of WIP relative to how fast it finishes parts, every new part has to wait behind everything already in the queue. More WIP is literally more wait. That is the mathematical reason a plant can pile up inventory and get slower at the same time.
The batch effect stacks on top. In a batch of 50, the first part finishes an operation and then waits for the other 49 before the whole lot moves, so it carries 49 parts' worth of wait even though its own work took seconds. Shrinking the transfer batch attacks that wait directly, which is the core mechanic behind one-piece flow and kanban pull. The result shows up as higher throughput for the same equipment, because parts stop queuing behind their own batch.
The relationship is linear and unforgiving. Hold the throughput steady and double the WIP, and you double the lead time, every extra job in the system is extra wait for everything behind it. That is why "release more work to keep people busy" so often backfires: it raises WIP, lengthens every queue, and slows delivery while the plant looks busier than ever. The line below is not a curve you can argue with; it is arithmetic.
How do you reduce wait time?
Work the wait in order, cheapest lever first. You do not need capital to move most of this; you need to stop releasing work faster than the constraint can take it, and stop moving parts in big lumps.
- Measure the wait, not just the work. Follow one part and timestamp the gaps between operations. You cannot cut a queue you never put a number on.
- Cap what you release. Set a work-in-process limit so the floor cannot pile up jobs faster than they clear. Less WIP is directly less queue wait, by Little's Law. See work-in-process reduction.
- Shrink the transfer batch. Move parts in tens instead of fifties. This attacks batch wait with no new equipment and usually no new labor.
- Balance the line to the constraint. Wait piles up wherever an upstream step runs faster than the one it feeds. Match rates so nobody outruns the slow step.
- Move inspection to the source. Inspecting at the operation that made the part removes the disposition queue and the rework loop at once. The customer never ordered the wait for a sign-off.
- Shorten the physical route. Put sequential cells next to each other. Every foot of travel is a chance to wait for a mover.
- Pre-authorize routine releases. Standing rules for normal holds mean parts do not wait on a person to notice them. Reserve human review for the exceptions.
What is batch delay, and how do you cut it?
Batch delay is the wait a part accrues while the rest of its lot is processed. It is the single most under-measured wait because it hides inside "normal" batch running. A part that takes 30 seconds at a step but travels in a lot of 100 can carry nearly 50 minutes of batch delay at that step alone, before any queue wait is added.
The fix is to separate the transfer batch from the process batch. You may still set up the machine to run 100 (the process batch), but you move parts downstream in tens (the transfer batch) so they start the next operation without waiting for the whole lot. Combined with faster changeovers through SMED this lets you shrink batches without losing capacity to setup time, the usual objection to smaller lots.
Batch wait also compounds across a routing. A part that travels through six operations in lots of 100 pays batch delay at each one, so the total wait is six stacked lot-times, not one. That compounding is why plants that move to smaller transfer batches often see lead time fall by more than they expected, they are draining the same waste at every step of the route at once, not just at the first machine.
By the numbers
The scale of the opportunity is national, not anecdotal. U.S. manufacturers carried an inventories-to-shipments ratio of 1.47 in May 2026 on roughly $962.0 billion of total inventory, per the U.S. Census Bureau's M3 report about a month and a half of shipments held as stock, most of it waiting rather than being worked. That squares with the lean finding, reflected in the U.S. EPA's lean methods guidance that value-added time is a small share of total lead time before improvement. Wait, in other words, is not a rounding error you optimize last. It is the main event.
How do you measure wait-time reduction?
Track lead time and its split, not a single average. The honest scorecard is total lead time, the value-added share of it (process cycle efficiency), and WIP at the worst queue. When a fix works, lead time drops and PCE rises for the same work content, and by Little's Law, WIP at the queue falls too. Fold those into your manufacturing KPIs so wait shows up next to output, and put the same numbers on the timeline of your value stream map so the whole team sees where the hours went.
Sustaining it is a capture problem. Wait hides because systems log transactions, scanned in, scanned out, not the hours between them. Plants like CLS replaced paper logging with real-time capture, so queues are visible while they are forming instead of reconstructed after the shift. If you want to size the dollars behind your losses before you start, run your line through a free OEE calculator.