Pattern/Systems Thinking/No. 1030
Theory of Constraints
The theory of constraints is Eliyahu M. Goldratt’s management approach to improving systems by addressing the few limits that restrict their goals. It focuses work on the current constraint, or bottleneck, rather than on making each part faster, then repeats as the limit moves.
- Evidence
- Useful, modest evidence
- Read
- 6 min
- Links
- 9 connections
01You've seen this when…
- at work
Developers finish features faster after hiring two more engineers. The release queue keeps growing because one security reviewer must approve every change.
- in life
You chop vegetables early and recruit someone to set the table. Dinner still runs late because three dishes need the same oven at different temperatures.
- out in the world
A city puts permit applications online. Submissions arrive instantly, but approvals still take six weeks because the same small team reviews every file.
02The idea
Everyone can work faster without the system delivering more. If every finished order needs the same testing bench, speeding up assembly mostly creates a larger pile beside it.
The theory of constraints starts there: a system’s ability to achieve its goal is limited by one or a few constraints. Find what currently limits the result, use it better, and organize the surrounding work to support it. Only then consider adding capacity. When the limit moves, repeat.
A constraint can be a slow machine, scarce expertise, customer demand, a supplier, or a rule that prevents available capacity from being used. Start by naming the goal. More units produced, more profitable sales, and shorter waiting times are different goals and may have different constraints.
A bottleneck is a limiting point in a process. The theory of constraints is a broader management approach: it tells you how to coordinate departments’ improvements around that limit so their efficiency serves the whole system.
The central distinction is between improving a part and improving the whole. A faster part matters only if it changes the result you care about.
03Why it happens
- Required steps depend on one another. In a simple production line, every item must pass through every stage. The stage with the lowest effective capacity caps the line’s sustained output, however fast the others work.
- Extra work accumulates before the limit. If assembly produces 40 items a day and testing clears 20, the excess stays unfinished as work in progress. It takes up space and ties up money while demanding attention.
- Variation makes spare capacity valuable. Breakdowns, uneven arrivals, and difficult cases interrupt flow. Keeping every resource busy leaves little room to recover. Queueing theory explains why waiting can rise sharply near full utilization.
- Local targets pull work in the wrong direction. A department rewarded for its own output may keep producing things the next stage cannot handle. This is a local-versus-global optimization problem: each part improves its score while the overall result stalls.
04A worked example
In this invented example, a workshop makes identical sensors. Every sensor moves from assembly to calibration before reaching packing. At their effective capacities, assembly can handle 40 sensors a day, while calibration can process 20 before passing them to packing, which can handle 50 a day. Customers want 30 sensors a day.
The manager buys assembly tools that raise its capacity to 60. Calibration stays at 20.
What it looks like A successful productivity investment. Assembly can now finish 50% more sensors, and its performance report improves.
What’s actually going on The workshop still ships at most 20 sensors a day. If assembly uses all its new capacity, unfinished sensors accumulate even faster. Calibration remains the limit on deliveries after the investment. Assembly improvements might reduce costs or defects. Extra assembly capacity alone leaves this line’s output capped by calibration.
What would have helped First protect calibration time. Prepare the bench’s inputs so fixtures are ready before it needs them and necessary parts are available, and catch avoidable defects before they consume calibration time. Limit assembly releases to support calibration and prevent work from flooding it. If calibration still limits profitable sales, evaluate extra calibration capacity.
At a calibration capacity of 30 a day, the workshop could meet current demand, assuming the other stages remain available. Beyond that, customer demand becomes the constraint. Buying more equipment would still leave sales dependent on demand.
05How to spot it
These are clues, not proof. A queue may reflect batching or priority rules rather than a persistent capacity shortage.
06What to do about it
Goldratt’s approach is commonly organized into five focusing steps:
- Identify the current constraint. Define the system’s goal, then trace what prevents more of it. Follow actual work and waiting times to test whether the busiest person is the constraint.
- Exploit the constraint. Get more value from what you already have. Remove avoidable interruptions and prepare inputs so you can reserve scarce time for work that advances the goal. Here, exploit means using available capacity effectively while protecting the people involved from exhaustion.
- Subordinate other work to it. Adjust schedules, priorities, and releases to support the constraint. Allow some nonconstrained resources to be idle. Keep a deliberate protective buffer of queued work where interruptions would otherwise starve the constraint.
- Elevate the constraint. If the limit still matters, consider investment: training, equipment, additional staff, outsourcing, or changing a restrictive policy. Compare the expected improvement with its full cost.
- Repeat when the constraint moves. Check what now limits the goal. Shift improvement efforts to the current constraint, even when the program for yesterday’s bottleneck still has momentum.
Measure the system’s result alongside its unfinished work. If inventory rises while completed, useful output holds steady or falls, the extra activity may be adding to the backlog. Little’s law connects the amount of work in progress to the completion rate and time in the system.
07Where it doesn’t explain the whole system
The simple version works best when the goal is clear and work follows a reasonably stable route. Hospitals often handle different kinds of work to serve several goals, under constraints that shift by hour or case—a challenge that software teams share with public agencies. A single permanent bottleneck may be the wrong model.
Nor does the approach make every improvement outside the constraint pointless. Safety, quality, resilience, and cost reductions can matter even when they don’t raise output. Slack resources may protect performance rather than represent waste.
TOC measures throughput in money: sales revenue minus totally variable costs. Faster production of unsold goods produces no financial gain. And choosing the goal remains a judgment: maximizing sales alone leaves open what is owed to workers and customers and how obligations to the public should be met.
08Roots
Before becoming known for a business novel, Eliyahu Goldratt was an Israeli physicist working on factory-scheduling software. Manufacturing gave him a recurring puzzle: managers could keep individual machines busy and still fail to deliver orders. The measurements encouraged activity at each workstation, while the problem lived in the connections between them.
Goldratt and Jeff Cox made that conflict readable through fiction in The Goal, first published in 1984. Its plant manager, Alex Rogo, faces a threatened factory closure. On a hiking trip, he notices a line of children repeatedly stretching out behind a slow walker, Herbie. The group’s shared pace becomes the focus when he moves Herbie to the front and redistributes his backpack. The hike turns factory flow into something a reader can picture.
Goldratt developed the factory ideas into a wider management approach covering constraints, measures, policies, and ongoing improvement. Managers applied it to distribution and services and used it to guide projects. Its lasting appeal is the change in attention: assess whether keeping a resource busy helps the whole system do better, and investigate what limits that result.
09How solid is this?
Capacity limits and queueing effects are well established. Published TOC case reviews report substantial improvements, but largely describe successful applications rather than controlled comparisons. The broader claim that every manageable system has only a few decisive constraints is less secure.
10Connections
- IncludesLiebig's Law of the Minimum, Work in Progress Limit, Bottleneck
- See also Diminishing Returns, Little's Law, Queueing Theory, Stocks and Flows, Local vs. Global Optima, Slack Resources
11Origin and sources
Eliyahu M. Goldratt, building on factory-scheduling work and introducing the core ideas widely through The Goal, coauthored with Jeff Cox (1984). He subsequently developed them into the theory of constraints management approach.
- [1]Goldratt, E. M., & Cox, J. (2004). The Goal: A Process of Ongoing Improvement (3rd rev. ed.). North River Press.
- [2]Mabin, V. J., & Balderstone, S. J. (2003). The performance of the theory of constraints methodology: Analysis and discussion of successful TOC applications. International Journal of Operations & Production Management, 23(6), 568–595.
- [3]Hopp, W. J., & Spearman, M. L. (2008). Factory Physics (3rd ed.). McGraw-Hill/Irwin.
Suggest an edit· Updated 2026-10-02