It usually starts quietly. On-time delivery slips from the mid-nineties to the high-eighties. A customer calls for status on an order that should have shipped last week. Someone starts a hot list. Within a month the hot list is the schedule, the morning meeting is an argument about which fire is biggest, and the plant is running Saturdays.
For a few weeks the overtime works. Then the late orders come back, the people are tired, premium freight is climbing, and the question from leadership gets sharper: why are we still late when we are working more hours than ever?
The honest answer is that overtime is a blanket solution to a specific problem. It adds hours everywhere, but only hours at one or two places in your operation actually control how much you ship. Until you know where those places are and what is really stealing time there, extra hours mostly create extra inventory in the wrong spots. This guide walks through how an experienced lean practitioner diagnoses late deliveries, and what to do about each cause.
Why “everyone is partly right” keeps you stuck
Ask five people why orders are late and you will get five answers. Sales says the schedule is unrealistic. Scheduling says production never hits the plan. Production says material shows up late and the machines keep going down. Quality says rework is eating capacity. Purchasing says the forecast was wrong.
Every one of them is describing something real. That is exactly the trap. When every theory is partly true, every department can defend its own position, and improvement effort gets spread thin across all of them. Nothing moves far enough to change the result.
The way out is to stop debating theories and start tracing evidence. Late orders leave a trail: where they were waiting, for how long, and why. Follow that trail and the argument usually ends within a couple of weeks.
The five real causes of late deliveries
In practice, almost every late-delivery problem traces back to some combination of five causes. Most plants have two of them driving the majority of the lateness.
1. Constraint capacity
Every operation has a constraint — the resource with the least capacity relative to the demand placed on it. The output of the whole plant is set by that one resource. An hour lost at the constraint is an hour lost for the entire plant; an hour saved anywhere else is, very often, a mirage.
At the constraint, time is lost in three ways: availability (breakdowns, waiting for material or operators, long changeovers), performance (running below rated speed, minor stops), and quality (making parts that are later scrapped or reworked).
This is where OEE belongs in the conversation. OEE is a useful measure of those three losses, but only at the constraint. A plant-wide average OEE is almost meaningless for delivery: a non-constraint running at 90% does not ship one extra order, and a comfortable average can hide the one asset that is actually starving the schedule. If your OEE reports look fine but you are still late, this is often why.
2. Flow and work-in-process
Lead time is driven far more by waiting than by working. When work moves in large batches and sits in queues between operations, the time from release to ship stretches out — and long lead times make everything harder to predict and easier to disrupt.
A useful rule from Little’s Law: lead time equals work-in-process divided by throughput. If you have three weeks of WIP on the floor and ship one week’s worth each week, your lead time is about three weeks, no matter how fast each individual operation runs. Releasing more work to “get ahead” only makes the queues longer.
The symptoms are familiar: piles of WIP in front of certain areas, constant expediting, jobs being pulled apart to rush a partial quantity, and a floor that has trouble telling which work is actually most urgent.
3. Schedule instability
Many plants do not have a production schedule so much as a daily suggestion. When priorities change every morning, setups are broken, batches are split, and yesterday’s plan becomes today’s scrap paper. Each change is made to rescue one order, and each change puts two or three others at risk.
Stable operations protect a frozen window — a short period (often a few days to a week at the constraint) in which the schedule does not change except for genuine emergencies. They also level the mix so that the constraint sees a steady, predictable load rather than waves of feast and famine.
4. Quality losses eating capacity
Scrap and rework are usually treated as a cost issue. For delivery, they are a capacity issue. A part scrapped after the constraint means constraint time was spent for nothing. Rework loops send work back through the same congested resources. Extra inspection added after an escape adds a queue of its own.
The question is not “what is our scrap rate?” but “how much constraint time are we losing to making things twice?”
5. Material availability
Sometimes the order was late because the material was late. Supplier delivery performance, planning parameters and purchasing lead times can all starve an otherwise healthy operation. This cause is real, but it is also the one most often blamed without evidence. Trace it like the others. If material truly is the driver, supplier performance and planning become the focus — work our sister company SupplySourceSync specializes in.
A step-by-step way out
Step 1: Stop blanket overtime until you know where hours help
This feels counterintuitive when you are behind, but blanket overtime is expensive and often counterproductive. Keep overtime targeted at the constraint and at the specific late orders, and stop running the whole plant extra hours to build inventory that is not needed yet.
Step 2: Trace your late orders
Pull the last 60 to 90 days of late orders. For each one, find where it was waiting when it went late, and why. You do not need perfect data; a spreadsheet with order number, due date, ship date, and “where it was stuck” is enough. Tag each order with one of the five causes. Patterns show up quickly.
Step 3: Find the real constraint
Walk the floor and look for the resource with the biggest pile of work in front of it and the least idle time. Confirm it with data: compare the load placed on each key resource to its demonstrated capacity. Be careful — the busiest-looking area is not always the constraint, and in some plants the constraint moves depending on product mix.
Step 4: Break down a week of constraint time
Study the constraint for a week and record where every hour goes: running at rate, running slow, changeover, waiting for material, waiting for an operator, down for maintenance, making scrap. This single exercise usually reveals more recoverable capacity than any capital request.
Step 5: Protect and improve the constraint
Common moves include running the constraint through breaks and shift changes, keeping a small buffer of ready work in front of it so it is never starved, cutting its changeover time with SMED, moving inspection so defects are caught before they consume constraint time, and giving it maintenance priority.
Step 6: Control WIP and stabilize the schedule
Release work to the floor at the pace the constraint can handle, not faster. Set a frozen window. Level the mix where possible. As WIP comes down, lead time shortens, the floor can see priorities clearly, and expediting drops.
Step 7: Run a daily delivery review
A short, standing daily review — ideally at a visual board — covering yesterday’s shipments versus plan, today’s at-risk orders, constraint performance, and the top blocker. Keep it to fifteen minutes and focus on removing obstacles, not assigning blame.
Choosing the right improvement method
| What the diagnosis shows | Typical method |
|---|---|
| A single, well-understood issue at the constraint (e.g., a recurring material shortage at one cell) | A3 problem solving |
| Changeover, layout, balance or flow problems in a defined area | Kaizen event |
| Chronic variation — scrap, cycle time or process instability with unclear root causes | Six Sigma / DMAIC project |
| Lateness spread across the whole value stream | Value stream mapping, then a prioritized set of the above |
Matching the method to the size of the problem matters. Using a months-long DMAIC project on a changeover problem wastes time; using a one-week Kaizen on a complex variation problem often produces a temporary fix.
The mistakes that keep plants late
- Improving everything at once. Effort spread across every area means no area improves enough to matter.
- Chasing plant-wide OEE. It feels productive and rarely moves delivery.
- Releasing more work to catch up. More WIP means longer lead times and more chaos.
- Letting the schedule change daily. Every rescue creates the next emergency.
- Treating premium freight as normal. It is a symptom that should be tracked and reduced.
- Buying capacity before exploiting the constraint. Expensive, and it often moves the problem rather than solving it.
What good looks like
When delivery performance is under control, the plant looks noticeably calmer. The constraint is known to everyone and protected. WIP levels are visible and limited. The schedule inside the frozen window rarely changes. The daily delivery review takes fifteen minutes and spends most of its time on prevention rather than recovery. Premium freight is an exception you can explain, not a monthly line item. And when demand changes, the team knows where to look first.
Just as important, once you have fixed one constraint, you know the constraint will move. Mature operations treat this as normal and simply go after the next one.
How QMSLean helps
We start with the evidence: tracing late orders, confirming the constraint, and separating the five causes with data. Then we choose the right method for each cause — A3, Kaizen or DMAIC — and work alongside your team to fix it and build the daily routines that keep delivery on track. If the root cause turns out to be supplier performance, we bring in our sister company SupplySourceSync rather than trying to solve a supplier problem from inside the plant.
If this sounds like your plant, see our situation page: “We’re missing ship dates, and overtime isn’t fixing it.” You may also find how to recover capacity before adding a shift or building and our complete guide to value stream mapping useful.
Tell us what’s going on
Describe your delivery situation — what you are seeing, what you have tried, and what is at stake — and a Lean practitioner will reply with how we would approach it. Request a Situation Review, try our free ROI Calculator, or contact us at [email protected].
Is this your situation?
See how QMSLean solves it — and request a Situation Review
Frequently Asked Questions
Why are we still missing ship dates when we are working overtime?
Overtime adds hours everywhere, but only the hours at your constraint increase output. Extra hours at non-constraints mostly build work-in-process, which lengthens lead time. Targeting improvement and overtime at the real constraint is what moves on-time delivery.
Does OEE tell us why deliveries are late?
Only if it is measured at the constraint. A plant-wide average OEE can look healthy while the one asset that paces output is losing significant time to downtime, changeovers or scrap.
How do we find our constraint?
Look for the resource with the largest queue of work in front of it and the least idle time, then confirm it by comparing the load placed on each key resource with its demonstrated capacity. The constraint can shift with product mix, so check it periodically.
What is a frozen schedule window?
A short period — often a few days to a week at the constraint — during which the production schedule does not change except for genuine emergencies. It protects setups and batches and stops daily reprioritization from creating new late orders.
How quickly can on-time delivery improve?
Once the constraint and the dominant causes are identified, focused improvement at the constraint often shows results within weeks. Structural changes such as WIP control and schedule leveling take longer but make the gain last.