# Theory of Constraints: Find the Bottleneck That Limits Everything

> Theory of constraints explained: Goldratt's five focusing steps for finding the bottleneck that limits any system, with business, personal and tech examples.

Canonical: https://mentarium.io/mental-models/theory-of-constraints.md

Mental model

6 min read By [Nemo](https://mentarium.io/about) Updated October 30, 2026

> The **theory of constraints** is the idea that every system’s output is limited by one constraint, its bottleneck, at a time. Developed by Eliyahu Goldratt, it says that improving anything other than the bottleneck adds no output, so you find the constraint, exploit it, subordinate everything else to it, then elevate it.

- **Category**: Mental model (operations and systems)
- **Origin**: Eliyahu Goldratt, The Goal, 1984
- **Best used for**: Deciding where effort will actually raise output
- **Related**: Pareto principle, feedback loops, leverage

## Ask yourself

- Where does work pile up and wait?
- If I made this step twice as fast, would total output rise?
- Is the bottleneck ever idle, and why?
- Has the constraint moved since I last looked?

## What is the theory of constraints?

The theory of constraints says that every system has one constraint at a time that limits its total output, and that the only way to get more output is to improve that constraint. Improving any other step makes you busier, not faster.

Picture a production line with four stations that can handle 100, 60, 90 and 120 units an hour. The line produces 60. Speeding up the 120-unit station to 200 changes nothing. Speeding up the 60-unit station to 80 raises the whole line to 80.

This sounds obvious once said, and that is the point. Most organizations still spread effort evenly, reward every department for being busy, and celebrate local improvements that never show up in results.

Constraints come in three broad types:

- **Physical**: a machine, a person, a room, a budget.
- **Policy**: a rule, an approval step, a habit (“we only release on Fridays”).
- **Market**: not enough demand, so sales is the bottleneck.

Goldratt argued that policy constraints are the most common and the cheapest to fix.

## Where it comes from

Eliyahu M. Goldratt, a physicist, developed the theory after building production scheduling software in the late 1970s. He spread it through _The Goal_ (1984), written with Jeff Cox: a novel about a plant manager, Alex Rogo, who has three months to save his factory.

The book’s best-known scene is a scout hike. The troop can only move as fast as its slowest boy, Herbie, and the gaps in the line keep growing. Alex puts Herbie at the front, redistributes his heavy pack, and the whole troop speeds up. The lesson: in a chain of dependent steps with ordinary variation, the slowest link sets the pace, and delays pile up rather than average out.

Goldratt later extended the ideas to project management in _Critical Chain_ (1997) and to scheduling with a method called drum-buffer-rope, where the bottleneck sets the rhythm (drum), a buffer of work protects it from starving, and a signal (rope) keeps upstream steps from flooding the system.

## How to use the theory of constraints

Use Goldratt’s five focusing steps.

1. **Identify the constraint.** Look for where work waits. Queues, piles, backlogs and people asking “is it done yet?” pile up in front of the bottleneck. Ask: if this step doubled its speed, would final output rise?
2. **Exploit the constraint.** Get the most from it as it is, without spending. Make sure it never sits idle, never works on defective input and never does tasks someone else could do.
3. **Subordinate everything else.** Set the pace of every other step to the constraint. Upstream steps should not overproduce; extra work just becomes inventory. This is the hardest step because it asks other teams to be less “efficient”.
4. **Elevate the constraint.** Only now add capacity: hire, buy, automate, redesign. This is where [leverage](https://mentarium.io/mental-models/leverage) belongs.
5. **Repeat, and watch for inertia.** Once a constraint is broken, another step becomes the limit. Go back to step 1. Goldratt warned that old rules written for the old bottleneck often become the new one.

The [Pareto principle](https://mentarium.io/mental-models/pareto-principle) says a few causes drive most results. The theory of constraints goes further for flows: in a chain, one step drives the result.

A quick way to test whether you found the real constraint: imagine giving it one free hour or one extra unit of capacity, and trace what happens to final output. If output rises, you found it. If the extra capacity just piles up in front of the next step, the constraint is somewhere else.

It also helps to measure in throughput, not activity. Goldratt defined throughput as the rate at which the system generates money through sales, not through production. Finished goods sitting in a warehouse are not throughput. Neither are features nobody uses.

 Mentarium Thinking Skills

Make Claude and ChatGPT think with mental models

12 ready skills: first principles, inversion, pre-mortem, second-order, a bias audit, a red team and more. Install once, then ask for the model by name.

 [See Thinking Skills · $29](https://mentarium.io/shop/thinking-skills)

## 3 real-world examples

### Business

Say a small agency has plenty of leads, three designers with free time, and one founder who approves every proposal. Proposals wait a week for review, and deals go cold. Hiring a fourth designer adds nothing. The constraint is the founder’s review. Exploit: block two hours a day for reviews only. Subordinate: designers send complete, checked drafts so reviews are fast. Elevate: write approval criteria so a senior designer can sign off on small proposals.

### Personal life

Imagine you want to write a book and keep buying courses, outlining and tweaking your setup. Output stays at zero. The constraint is drafting time, maybe 30 focused minutes a day that keep getting eaten. Exploit: protect those minutes first thing, phone in another room. Subordinate: research and editing happen only to feed the next draft session. The outline does not need a fifth version.

### Tech

A software team ships slowly. Developers want faster laptops, but tickets actually wait days for code review by two senior engineers. That is the constraint. Exploit: reviews first each morning. Subordinate: smaller pull requests with clear descriptions. Elevate: train more reviewers. When review speeds up, the constraint may move to testing or deployment. Go back to step 1.

## Common mistakes

- **Optimizing the loudest problem.** The step that complains most is not always the bottleneck. Follow the queues, not the noise.
- **Keeping everyone busy.** Full utilization of non-bottlenecks creates piles of unfinished work and hides the real constraint.
- **Elevating before exploiting.** Buying capacity is expensive; most constraints have slack you can recover for free.
- **Forgetting the constraint moves.** Fixes work, and then the limit shifts. Check again.
- **Ignoring policy constraints.** Rules and approvals often limit flow more than any machine.
- **Measuring local efficiency.** Rewarding each department for its own utilization pushes everyone to overproduce and starves the real constraint of attention.

## Theory of constraints vs Pareto principle

|  | Theory of constraints | Pareto principle |
|---|---|---|
| Core claim | One constraint limits system output | A few causes drive most effects |
| Best for | Flows of dependent steps | Any distribution of causes and effects |
| Question it asks | Which step limits the whole? | Which inputs matter most? |
| Method | Five focusing steps | Rank and focus on the top few |
| Origin | Goldratt, The Goal, 1984 | Vilfredo Pareto, generalized by Joseph Juran |
| Typical mistake | Fixing non-bottlenecks | Treating 80/20 as an exact ratio |

## The Mentarium take

The theory of constraints is the best antidote we know to effort spread thin. It counters the [availability heuristic](https://mentarium.io/biases/availability-heuristic): we fix what is vivid and recent, the angry email or the visible delay, instead of the step that quietly limits everything. Goldratt’s question, “where does work wait?”, replaces impressions with evidence.

Our favorite use is personal. Most people’s productivity has a single constraint at any given time: energy, focus hours, a decision they keep postponing, a skill they lack. Find it with [five whys](https://mentarium.io/mental-models/five-whys) and the queues in your own life (the inbox, the unfinished drafts, the pile of half-done chores), then follow the five steps. A Model Stack that works well here is Theory of constraints + Pareto principle + [Feedback loops](https://mentarium.io/mental-models/feedback-loops): find the limit, focus on the few changes that move it, and watch what loops your fix sets off.

And remember the inertia warning. The rules you made to fix last year’s bottleneck can be this year’s. Review them on purpose.

**Counters** [Availability Heuristic](https://mentarium.io/biases/availability-heuristic)

## FAQ

### What are the five focusing steps of the theory of constraints?

Identify the constraint, exploit it (get the most out of it as it is), subordinate everything else to it, elevate it (add capacity), and when it is broken go back to step one without letting inertia become the new constraint.

### Who created the theory of constraints?

Eliyahu M. Goldratt, an Israeli physicist turned management thinker. He introduced it in The Goal (1984), a business novel written with Jeff Cox, and developed it in later books such as Critical Chain (1997).

### What is a bottleneck in simple terms?

A bottleneck is the slowest step in a process that everything else must pass through. Like the neck of a bottle, it sets the maximum flow no matter how wide the rest is.

### Is the theory of constraints only for factories?

No. It started in manufacturing but applies to any flow of work: software delivery, sales pipelines, hiring, writing a book, or a personal schedule. Anywhere work moves through steps, one step limits the rest.

### What is the difference between the theory of constraints and lean?

Lean focuses on removing waste across the whole process. The theory of constraints focuses improvement on the single step limiting output. Many organizations use both: the constraint shows where to look first, and lean tools help improve it.

## Related models & biases

[Mental model Pareto Principle (80/20 Rule) Explained With Examples The Pareto principle, or 80/20 rule, is the observation that in many systems a small share of causes produces most of the effects: a few customers bring most of the revenue, a few bugs cause most of the crashes. The 80 and 20 are a rough shape, not a law.](https://mentarium.io/mental-models/pareto-principle) [Mental model Feedback Loops A feedback loop is a cycle in which a system's output comes back as an input and changes what happens next. Reinforcing loops amplify change (growth or collapse), and balancing loops resist it (stability). As a mental model, it explains behavior through structure instead of single causes.](https://mentarium.io/mental-models/feedback-loops) [Mental model Leverage Leverage is anything that multiplies the effect of your effort, so a small input produces a large output. As a mental model, it sorts your options by how much they amplify you: through other people's labor, capital, code or media. It also warns that leverage multiplies mistakes as much as good judgment.](https://mentarium.io/mental-models/leverage) [Mental model The 5 Whys The 5 Whys is a root cause analysis technique: you state a problem, ask why it happened, then ask why again of each answer, roughly five times, until you reach a cause you can fix at the process level. It was developed at Toyota, credited to Sakichi Toyoda, and spread through the Toyota Production System.](https://mentarium.io/mental-models/five-whys) [Cognitive bias Availability Heuristic The availability heuristic is a mental shortcut in which you judge how common or likely something is by how easily examples come to mind. Recent, vivid, emotional or heavily covered events are easy to recall, so they feel more frequent than they really are, while quiet, common risks feel rarer.](https://mentarium.io/biases/availability-heuristic)

 Mentarium Decision Coach · free

Use this with your AI

Connect the free Decision Coach to Claude or ChatGPT and it walks you through the Decision Loop, picks the right Model Stack and runs the Bias Radar on your reasoning.

 [Connect it free](https://mentarium.io/coach)

 Mentarium Decision Coach · free

Use this with your AI

 [Connect it free](https://mentarium.io/coach)