Lead time reduction is the practice of shortening the total time between an order and its delivery, mostly by cutting the waiting between operations rather than speeding up the operations themselves. In a typical plant, actual processing is only 5–10% of lead time; the other 90%-plus is material sitting in a queue.
That single fact reorders every improvement priority. If you chase faster machines and better methods, you are optimizing the 10% and ignoring the 90%. The fastest, cheapest lead time gains almost always come from attacking queues and batch sizes, the reasons product waits, not from shaving seconds off a cycle. This guide shows where lead time hides, why waiting dominates, the steps that actually move it, and what Little's Law says about the whole thing.
Where Does Lead Time Actually Go?
Lead time goes almost entirely to waiting: material sitting between operations, in queues, in staging, in inspection holds, and in the warehouse. Studies of manufacturing flow consistently find that value-adding processing is a small slice of the total, often 5–10%, while the remainder is pure wait. A part that takes 40 minutes of actual work can spend three days in the building.
This is why measuring lead time by adding up cycle times gives a number that is wildly optimistic. Your routing might sum to two hours of work; your actual order-to-ship is two weeks. The difference is the waiting your cycle-time data never sees, because cycle time measures work at a station and lead time measures the whole journey, including all the gaps. If you have not separated these two, start with lead time vs. cycle time confusing them is the most common reason improvement effort lands on the wrong 10%.
Why Attack Queue and Batch Size Before Processing Speed?
Because batch size is the biggest single driver of the waiting you are trying to kill. Run parts in a batch and every part must wait for the whole batch to finish before any of them move to the next operation. A batch of 200 parts at 2 minutes each takes 400 minutes to complete, so the first finished part sits idle for roughly 6.7 hours before the batch releases downstream. Cut that batch to 20 and the wait drops by an order of magnitude, with no change to how fast the machine runs.
Queues compound the effect. As a resource gets busier, the line waiting for it grows non-linearly: wait time climbs sharply as utilization approaches 100%, which is the mathematical reason a plant that loads every machine to the limit ends up with long, unpredictable lead times. Deliberate slack at non-constraint operations feels wasteful and actually shortens lead time, because it keeps queues from exploding. This is the same logic behind theory of constraints: protect the bottleneck, but do not choke the rest of the line with work it cannot process yet.
How Do You Reduce Lead Time?
You reduce lead time by systematically removing the reasons product waits, in order of impact. The sequence that works:
- Map the whole flow and mark the waits. Walk the order from receipt to ship and record every place product sits. A value-stream map that shows process time versus wait time at each step makes the 90% visible and undeniable.
- Shrink transfer batches first. The cheapest lever. Move product in smaller quantities between operations so downstream steps start sooner. You do not need to change the production batch to shrink the transfer batch.
- Cut changeover time so small batches are affordable. Big batches usually exist to amortize long setups. Attack the setup with SMED world-class programs cut changeover time by 70–90%, and small batches stop being expensive.
- Cap work-in-process and release by capacity. Stop dumping orders onto the floor. Release work at the rate the constraint can absorb it; a controlled WIP cap keeps queues short and lead time predictable.
- Attack the biggest queues at the constraint. Protect the bottleneck with a small buffer, but relieve every other queue. A minute saved at a non-constraint does nothing for lead time; a queue killed ahead of the constraint shortens the whole order.
- Remove hidden holds. Inspection queues, approval waits, staging for a truck, and quality holds are lead time too. A part waiting in a QA hold is as late as a part waiting at a machine.
- Level the schedule. Erratic release creates waves of queue. Smoothing the mix, takt-paced release and level loading, keeps lead time steady instead of lurching between rush and idle.
None of these speed up a single machine. They all attack waiting, which is where the time is. Ranked by the return you get for the effort, the levers tend to fall out like this:
| Lever | What it attacks | Effort | Typical impact on lead time |
|---|---|---|---|
| Shrink transfer batches | Batch wait between operations | Low | Large |
| Cap WIP / release by capacity | Queue length everywhere | Low–medium | Large |
| Reduce changeover (SMED) | The reason batches are big | Medium | Large (enabler) |
| Level the schedule | Waves of queue from erratic release | Medium | Moderate |
| Remove inspection / approval holds | Hidden non-machine waits | Low–medium | Moderate |
| Speed up a machine | The 5–10% that is processing | High | Small |
The pattern is consistent: the cheap moves attack waiting and pay off big, while the expensive move, buying speed, attacks the smallest slice of the total. Start at the top of that table and work down. Most plants exhaust the low-effort, high-impact rows before they ever need to consider capital.
What Does Little's Law Tell You About Lead Time?
Little's Law says lead time equals work-in-process divided by throughput, so if you want shorter lead time and cannot raise throughput, you must cut WIP. It is one of the most useful equations on the floor because it needs no stopwatch: count the jobs in the system and know the completion rate, and you can estimate lead time directly. The formal treatment is in our Little's Law explainer but the practical reading is blunt: piles of WIP are piles of lead time.
Run the arithmetic. A cell holding 500 units of WIP that finishes 100 units a day has a 5-day lead time, by Little's Law, no matter how fast any individual machine runs. Halve the WIP to 250 and, at the same throughput, lead time falls to 2.5 days. That is why capping WIP is not bookkeeping; it is the most direct lever you have on how long a customer waits. And it is why building ahead, inflating WIP to keep people busy silently lengthens lead time for every order behind the pile.
Lead time reduction: the reference numbers
The proportions that make queue-first the right strategy are well documented:
- Value-adding processing is commonly only 5–10% of total manufacturing lead time materials often spend 90–95% of their time waiting between operations. This is the recurring finding behind lean flow work and the reason waiting is the largest target.
- SMED can cut setup times by 70–90% the enabler that makes small batches economical (the technique Shigeo Shingo formalized under Single-Minute Exchange of Die).
- Capacity-based work release is reported to cut lead times 30–50% within 3–6 months with no change to processing speed or added capacity, pure queue reduction.
- U.S. manufacturing runs near 75.7–76.3% capacity utilization per the Federal Reserve's G.17 release (Federal Reserve, Industrial Production and Capacity Utilization), which leaves room to add non-constraint slack that shortens lead time rather than lengthening it.
How Much Lead Time Can You Realistically Cut?
Plants that attack queues and batch sizes commonly report 30–50% lead time reductions within a few months, without buying faster equipment. The gains are large precisely because the starting point is so wasteful: when 90% of lead time is waiting, there is enormous room to compress it before you ever touch a machine. The first pass, shrinking transfer batches and capping WIP, is usually the biggest and cheapest.
A word of realism: lead time reduction is not a one-time project. Queues regrow the moment discipline slips, batches creep back up when a setup gets neglected, and WIP piles return the first time a supervisor decides to "keep the line busy" through a slow order week. The plants that hold their gains treat lead time as a standing metric on the same board as safety and quality, reviewed every week, with the biggest current queue named as a problem to solve. Improvement events shorten lead time; only a management routine keeps it short. Tie the number to a visible owner and a live view of the floor, and the ratchet holds.
The durable version of this depends on seeing the waits as they happen instead of reconstructing them after the fact. When WIP levels, queue lengths, and hold times are captured from machine signals and the floor's own tablets, the queues that grow overnight are visible before they blow out the schedule, and you can tell whether a lead-time gain held or quietly crept back. That is the approach Harmony takes when it connects the floor into one operational layer: WIP and flow are computed from source, not tallied from paper at week-end, with no rip-and-replace of the equipment. The plant in our CLS case study runs its shortening effort off live floor data, which is what keeps the gains from eroding after the kaizen team goes home.