A line has the capacity to produce 100 units per hour. It only produces 62. Where did the other 38 go? This is the gap that costs manufacturers millions every year — and it starts with confusing two very different concepts. Capacity is what your system could produce. Throughput is what it actually produces. Understanding the difference, and what drives it, is one of the most valuable things you can do to improve manufacturing performance.
1. The Gap Nobody Talks About
Most improvement efforts focus on increasing capacity — buying equipment, adding shifts, increasing line speed. But in most facilities, the problem is not capacity. The problem is that existing capacity is never fully converted into output. Bottlenecks, downtime, changeovers, and line imbalance silently consume between 20% and 50% of available production time. The chart below shows what that gap looks like in practice.
A line rated at 100 units/hr producing only 62 — 38 units of potential output silently lost every hour.
2. What Is Throughput?
Throughput is the actual rate at which a system produces and delivers output. It is not a theoretical number — it is measured from real production data. Every minute of downtime, every slow changeover, and every bottleneck constraint directly reduces throughput. It is the single most honest measure of production performance.
Throughput Formula
Example: 480 units produced in an 8-hour shift = 60 units/hr throughput
The key insight is that throughput is a system property, not a machine property. A single slow station, a starved feeder, or a quality problem downstream will reduce the throughput of the entire line — even if every other machine is running perfectly. The animation below shows how a single slow stage controls what exits the system.
Units flow freely until they reach the bottleneck — the slowest stage controls what exits the entire line.
3. What Is Capacity?
Capacity is the maximum possible output a process, machine, or system is theoretically capable of producing. It is typically defined by the equipment’s rated speed, available time, and design specifications — assuming no losses. Capacity is a ceiling, not a guarantee. It represents potential, not performance.
Capacity Formula
Example: 100 units/hr × 8 hours = 800 units potential capacity per shift
The gap between a machine’s nameplate rating and its actual output is almost always larger than people expect. Downtime, speed losses, minor stoppages, and quality defects all chip away at it. The visual below shows a common scenario — a machine rated at 100 units/hr that actually delivers 67.
Rated capacity vs. actual output — unavoidable losses always create a gap between the two.
4. The Key Difference
The simplest way to remember the distinction: capacity is what you could do, throughput is what you do. Capacity is defined by design — it is relatively fixed and determined by your equipment and available time. Throughput is defined by reality — it is variable, driven entirely by your constraints, losses, and how well you manage flow through the system.
| Feature | Capacity | Throughput |
|---|---|---|
| Definition | Maximum possible output under ideal conditions | Actual output delivered over a time period |
| Driven By | Equipment design, rated speed, available time | Bottlenecks, downtime, changeovers, quality |
| Stability | Relatively fixed | Variable — changes shift to shift |
| Measured By | Design spec / nameplate rating | Real production data from ERP or counters |
| Can You Improve It? | Only through capital investment | Yes — by removing constraints and waste |
| Lean Focus | Secondary — capacity is rarely the issue | Primary — throughput improvement is the goal |
Notice the crucial implication in that last row. In most manufacturing environments, the limiting factor is not insufficient capacity — it is the failure to convert existing capacity into actual output. This means buying more equipment, running extra shifts, or increasing line speed will not solve the problem unless the underlying constraints are addressed first.
6. Example 1 — The Simple Production Line
Consider a three-stage production line. Process A cuts at 100 units/hr, Process B mills at 80 units/hr, and Process C finishes at 120 units/hr. What is the system’s throughput? The answer is 80 units/hr — dictated entirely by Process B, regardless of what the other stations can do. Designing A and C to run faster than B provides no benefit. That excess capacity is wasted.
Three-process line: throughput = min(100, 80, 120) = 80 units/hr. Process B controls the system.
Increasing Process A or C beyond the bottleneck rate does not increase throughput. It only creates WIP inventory piling up in front of Process B and idle time after it.
7. Example 2 — The Bottleneck Constraint
A single CNC machine in a fabrication shop can only process 40 components per hour due to high cycle times and frequent tool changes. Every other station in the line is capable of 90–110 units/hr. Despite all that available capacity, the entire factory’s output is capped at 40 units/hr. The constraint — not the facility’s capacity — defines what gets shipped to the customer.
Without Bottleneck Management
- WIP stacks up before the CNC machine
- Downstream stations sit idle and starved
- Output: 40 units/hr despite 90+ elsewhere
- Management buys a second upstream machine (wrong fix)
With Bottleneck Management
- CNC machine runs 100% of available time
- Upstream feeds only what CNC needs (no excess WIP)
- All improvement effort focused on the constraint
- Output increases as constraint improves
The constraint acts like a funnel — the narrowest point controls what flows through the entire system, regardless of upstream or downstream capacity.
8. Example 3 — The Idle Capacity Trap
A packaging line has four machines installed, each capable of 120 units/hr — giving an apparent capacity of 480 units/hr. But three of the machines sit idle for 60% of each shift due to a starved feeder at the start of the line. The fourth machine runs continuously but has nothing to process. Installed capacity is 480. Actual throughput is 48. This is the idle capacity trap — more machines do not equal more output when the constraint is upstream of them.
Left: high capacity, low throughput — idle machines waiting for work. Right: balanced flow — same machines all contributing to output.
The fix here is not to add a fifth machine. It is to fix the feeder that starves the existing four. Fixing flow delivers more output than adding capacity. This is a principle that runs through lean manufacturing, the Theory of Constraints, and every well-run production system.
9. Example 4 — The Changeover Impact
A high-mix production line runs 12 different SKUs per shift with average changeover times of 45 minutes each. With eight changeovers per shift, that is 360 minutes of the 480-minute shift consumed by non-productive setup time. Only 120 minutes — 25% of the shift — is actual running time. Installed capacity might be 200 units/hr, but actual throughput is under 50. The constraint is not equipment speed. It is changeover frequency and duration.
Before vs. after SMED-style changeover reduction — cutting setup time directly converts into production time and higher throughput.
Applying SMED (Single Minute Exchange of Die) principles to reduce changeovers from 45 minutes to 12 minutes on this line adds over 250 minutes of productive time per shift. That is more impactful than any equipment upgrade.
10. Why Capacity Never Equals Output
If capacity is the ceiling and throughput is reality, what fills the space between them? There are five primary categories of loss that consume the gap between what your system could produce and what it actually does. Understanding each one is the starting point for targeted improvement.
Bottleneck Constraint
The single slowest process in any line sets the throughput ceiling for the entire system. Until the constraint is elevated, all other improvements are secondary.
Unplanned Downtime
Breakdowns, minor stoppages, and unexpected machine failures remove large blocks of productive time. Industry average: 15–25% of available time lost to unplanned stops.
Changeover Time
Every minute spent setting up for the next product is a minute not producing output. High-mix lines are especially vulnerable to this loss.
Line Imbalance
When workloads across stations are unequal, faster stations sit idle waiting for the slowest. This mismatch creates structural waste that limits throughput even with no breakdowns.
Speed & Quality Losses
Running below rated speed, producing scrap, and reworking defects all reduce effective throughput below even the adjusted capacity figure.
Material Starvation
Delayed kitting, forklift wait times, and material handling inefficiencies regularly stop lines that are mechanically capable of running. Supply flow is part of the throughput equation.
The chart below shows how these losses break down in a typical manufacturing scenario, consuming the 38 units/hr gap between designed capacity and actual throughput.
Breakdown of the 38-unit gap between capacity and throughput — bottleneck constraint is typically the largest single loss.
11. How to Improve Throughput
Improving throughput requires a disciplined, prioritised approach. Random improvement projects that ignore the bottleneck will not move the dial on output. The five-step method below, rooted in Goldratt’s Theory of Constraints, gives a logical sequence for attacking the gap.
Use production data, WIP levels, and operator observations to identify the single process or resource that limits system throughput. Look for where inventory builds up, where people are always waiting, and which machine is always running at 100% while others cycle on and off.Identify the Constraint
Find what limits the system
Before changing anything, ensure the bottleneck is being used as productively as possible. Eliminate preventable stoppages at the constraint, ensure it is never starved of material, and prioritise its schedule above all other stations. Every minute of constraint time lost is a minute of system throughput lost.Exploit the Constraint
Get maximum output from it now
Adjust the pace, scheduling, and priorities of all other stations so they exist to feed and support the constraint — not to maximise their own utilisation. This is counterintuitive, but running upstream processes below capacity to prevent WIP overflow is often the correct action.Subordinate Everything Else
Align the whole system to support it
If steps 1–3 are not enough, invest to increase the constraint’s capacity: add a parallel machine, reduce cycle times through engineering, apply SMED to reduce downtime, or cross-train additional operators. Only do this after fully exploiting what you already have.Elevate the Constraint
Invest to break the limit
Once you elevate one constraint, a new one will emerge elsewhere in the system. Return to step 1 and start the cycle again. Throughput improvement is an ongoing process, not a one-time project. Each cycle raises the performance ceiling.Repeat — Find the New Constraint
The bottleneck will move
Before vs. after: addressing the bottleneck and reducing downtime delivers a 47% throughput gain without adding any capital equipment.
12. Common Mistakes
Most organisations make the same predictable errors when trying to improve output. These mistakes are expensive because they consume improvement effort and capital without moving the throughput needle. Recognise them before you fall into them.
Chasing Capacity Instead of Throughput
Adding machines, increasing headcount, and running overtime when the problem is a bottleneck that none of those investments will fix. Capacity is not the limiting factor in most manufacturing environments.
Maximising Every Machine's Utilisation
Measuring and rewarding individual machine efficiency creates overproduction, excess WIP, and local optimisation that harms the system. A machine running at 90% utilisation can be a symptom of poor flow, not good performance.
Ignoring the Bottleneck Location
Applying improvement effort to non-constraint processes. An optimised non-bottleneck remains a non-bottleneck — it does not change system throughput. This is one of the most common and costly wastes in continuous improvement programmes.
Confusing Throughput with Production Rate
Tracking how fast machines run rather than how many finished units exit the system. Parts moving fast between stations mean nothing if the bottleneck is not releasing more completed product. Only what ships counts as throughput.
The fundamental shift in thinking: stop optimising individual machines and start optimising system flow.
For a deeper dive into identifying and eliminating bottlenecks, see our Bottleneck Examples guide. To understand the waste that hides inside your process, read our Value-Add vs Waste guide. For root cause analysis when throughput problems appear, use the Root Cause Analysis framework.
Frequently Asked Questions
What is throughput in manufacturing?
Throughput is the actual rate at which a manufacturing system produces and delivers finished output over a given time period. It is calculated by dividing units produced by time elapsed, and is always limited by the slowest constraint in the process. Unlike capacity, throughput reflects real performance including all losses from downtime, bottlenecks, and changeovers.
What is the difference between capacity and throughput?
Capacity is the theoretical maximum output your system is designed to produce under ideal conditions. Throughput is the actual output achieved in practice. Capacity is fixed by design; throughput is variable and driven by constraints, downtime, and operational losses. In most manufacturing facilities, throughput is 60–80% of installed capacity due to these losses.
Why is throughput lower than capacity?
The gap between capacity and throughput is caused by five main factors: bottleneck constraints (the single slowest station limiting the whole line), unplanned downtime and machine failures, changeover and setup time consuming productive hours, line imbalance causing idle time at faster stations, and quality losses from scrap and rework. Fixing these losses is how manufacturers close the gap without buying new equipment.
How do you increase throughput?
The most effective approach is to follow the Theory of Constraints five steps: identify your bottleneck, exploit it fully before changing anything, subordinate all other activities to support it, then elevate its capacity if still needed. Reducing changeovers using SMED, improving planned maintenance to reduce unplanned downtime, and balancing line workloads are the three most common high-impact actions that improve throughput without capital expenditure.
What role does the bottleneck play in throughput?
The bottleneck controls everything. In any connected process, the single slowest station sets the maximum throughput for the entire system, regardless of how fast every other station runs. Improving non-bottleneck processes does not increase throughput — it creates WIP inventory and idle downstream capacity. All throughput improvement effort should be focused on the constraint first, until a new constraint emerges elsewhere.
