Manufacturing throughput glossary
Last updated: August 8, 2026
Plain-language definitions of the terms that come up when a team is trying to work out where a line is losing time. No jargon defined with more jargon. Where two terms are routinely confused, the difference is stated directly.
Throughput
The rate at which a process produces good output over a period of time, for example units per shift. Throughput is set by the bottleneck, not by the average speed of the line, which is why speeding up a non-bottleneck station changes nothing.
Cycle time
The time taken to complete one repetition of a process, measured from the start of one unit to the start of the next at the same point. Cycle time describes what the process actually does. Takt time describes what demand requires. Confusing the two is the most common terminology error in this area.
Takt time
The pace production must run at to meet demand, calculated as available production time divided by customer demand for that period. If cycle time exceeds takt time, the process cannot keep up with demand no matter how well it is running.
Lead time
The total elapsed time from an order or a unit entering the process to it leaving as finished output. Lead time includes all waiting and queuing, which is why it is usually far longer than the sum of the cycle times of the individual steps.
Bottleneck
The step or resource with the least capacity relative to demand, which therefore sets the throughput of the whole line. Work accumulates in front of it and the steps after it run starved. Improving a step that is not the bottleneck does not increase line output. The method for locating one is in how to find the bottleneck on a production line.
Constraint
The limiting factor on a system's output. In theory-of-constraints practice the constraint is what you improve first, because every other improvement is subordinate to it. On a production line the constraint and the bottleneck are usually the same thing.
OEE (Overall Equipment Effectiveness)
A composite measure of how much of the planned production time is genuinely productive, calculated as availability multiplied by performance multiplied by quality. OEE quantifies how much capacity is being lost at a machine or station. It does not identify which step within the work is losing it, which is the limit teams usually hit next.
Availability
The OEE component covering lost production time from stops, both planned and unplanned. Expressed as run time divided by planned production time.
Performance
The OEE component covering running slower than the ideal rate, including micro-stops and reduced speed. This is where the losses that manual tracking most often misses accumulate.
Quality (OEE component)
The OEE component covering units produced that do not meet specification first time, including anything that requires rework. Capacity spent producing a unit twice is capacity lost.
Changeover
The work required to switch a machine, cell, or line from producing one product to producing another. Changeover consumes capacity at the station performing it, so a station that changes over frequently can constrain a line more than a slower station that runs continuously.
SMED (Single-Minute Exchange of Die)
A structured method for reducing changeover time. It works by separating internal setup work, which can only happen while the machine is stopped, from external setup work, which can be prepared while it is still running, and then converting as much internal work to external as possible.
Standard work
The documented best-known method for performing a task, covering the sequence of steps, the timing, and the work-in-process required. Standard work is the baseline that improvement is measured against; without it, a change cannot be shown to be an improvement.
SOP (Standard Operating Procedure)
The written instruction describing how a specific task should be carried out. An SOP defines the intended method. It does not record how the work is actually being performed, which is the gap step-level measurement exists to close.
Time study
A manual industrial-engineering method in which a trained observer times the elements of a task, with a stopwatch or by reviewing video afterwards, and analyses the samples to establish a standard time. It is a long-established method and remains valid. It is also slow, and it produces timings rather than prioritized fixes. See time-study software alternatives for when it is still the right call.
AI procedure optimization
The use of AI to measure how a procedure is actually executed on the floor, step by step, identify where time and compliance are being lost, and return prioritized fixes. It automates the measurement, runs while the work runs, and produces recommended changes rather than only timings. The full definition is in what is AI procedure optimization.
Work-in-process (WIP)
Units that have entered the process but are not yet finished output. WIP accumulating in a particular place is one of the two physical signatures of a bottleneck, the other being idle time at the stations after it.
Line balancing
Distributing work across stations so that no single station holds substantially more work content than the others, bringing each station's cycle time close to takt time. A poorly balanced line creates a bottleneck by allocation rather than by capability.
Micro-stops
Short interruptions, typically seconds rather than minutes: a jam cleared by hand, a part reseated, a brief wait for a tool. Individually they are too small to be logged, so manual tracking reliably misses them, and collectively they can be a significant performance loss.
Value-added and non-value-added time
Value-added time is the portion of the work that changes the product in a way the customer is paying for. Non-value-added time covers everything else: walking, reaching, searching, waiting, inspecting, and moving material. Necessary non-value-added work, such as a required inspection, still counts as non-value-added.
Rework
Additional work performed on a unit to bring it up to specification after it has already been processed. Rework consumes capacity twice for a single unit of output, which is why it shows up as a throughput loss and not only as a quality problem.
Process variation
The spread in how long a step takes across repetitions, operators, and shifts. A step with wide variation indicates that the method is not controlling the work, and it is often more recoverable than the step with the highest average duration.
Continuous improvement (CI)
The ongoing practice of making small, measured improvements to a process, verifying each one, and repeating. Because relieving a constraint moves it elsewhere, improvement work is a loop rather than a one-off project.
Digital work instructions
Software that presents standard work to operators on a screen at the station, often with images, video, or confirmation prompts. Digital work instructions standardise how work should be performed; they do not measure how it is actually performed or where it is losing time. The comparison is in AI procedure optimization vs digital work instructions.
TurboProc Scope is an AI procedure-optimization tool for manufacturers. Capture a procedure on camera and the AI measures the timing and compliance of every step in real time, pinpoints the bottleneck, and returns prioritized fixes for the next shift. Get the app or start with the free procedure audit.