Structured edition
Theory of Constraints
by Eliyahu M. Goldratt
Faroa rebuilt the whole book as 12 concepts you read in order, at the depth you choose. The first concept is free to read in full - a 5-minute read.
Overview
Every system underperforms for one reason: it has a constraint holding the whole back.
Eliyahu Goldratt built a deceptively simple argument: most improvement efforts attack the wrong place. Fixing a non-constraint changes nothing meaningful.
Throughput beats efficiency.
The question is never how busy each part is. It is whether the system as a whole is moving toward its goal.
- Identify the one constraint limiting the system's output
- Subordinate every other decision to supporting that constraint
- Elevate the constraint, then find the next one
What lies ahead
These concepts move from identifying constraints to managing flow, measurement, and the thinking tools Goldratt developed to sustain continuous improvement.
What is inside
The Constraint Mindset
- 01The Goal Is Throughput, Not EfficiencyFind the one constraint that limits your system's output before optimizing anything else.Free, in full
- 02Every System Has One Binding ConstraintFind the step where work accumulates before doing anything else; that queue is your constraint.
- 03Local Optima Destroy Global PerformanceFind the single step that limits your system's output before changing anything else.
- 04Throughput, Inventory, and Operating ExpenseMeasure every decision by its effect on Throughput first, then Inventory, then Operating Expense, in that order.
Finding and Exploiting the Constraint
- 05The Five Focusing StepsFind the one step where work piles up and treat it as the whole system's priority.
- 06Subordinating Everything to the ConstraintSet your work release rate to match what the constraint can process, never what upstream resources can produce.
- 07Drum-Buffer-Rope SchedulingFind the true bottleneck first; every scheduling decision flows from that identification.
- 08Elevating the Constraint Without WasteConfirm the constraint with system output data before spending a single resource on elevation.
Thinking Tools and Organizational Change
- 09The Thinking Processes: Logic Trees for Any ProblemBefore proposing any solution, draw the cause-and-effect chain that produced the problem and validate every arrow with a colleague.
- 10Identifying the Core Conflict with the Evaporating CloudTrace each party's position back to its underlying need and then to the shared goal before attempting any solution.
- 11The Current Reality Tree and Root Cause DiagnosisMap undesirable effects backward through cause-and-effect before proposing any solution.
- 12Overcoming Resistance to ChangeSurface shared diagnosis before proposing any solution, agreement on the problem must come first.
Concept 01 of 12
The Goal Is Throughput, Not Efficiency
A factory running at full efficiency can still be losing money. The real goal is generating throughput, not keeping every resource busy.
When Busy Machines Starve the Business
Efficiency measures how hard each resource works. Throughput measures how much value actually leaves the system as finished output. These two metrics pull in opposite directions more often than most managers expect.
Goldratt's core claim is direct: the goal of a business is to make money, and the only activity that serves that goal is throughput, defined as the rate at which the system generates revenue through sales.
The Constraint Does All the Deciding
Every system has one step that limits the overall flow. That step is the constraint, and it determines throughput for the entire system, no matter how fast every other step runs.
Speeding up non-constraint steps does not raise throughput. It only builds inventory in front of the bottleneck and raises operating expense, both of which damage the business without producing additional revenue.
A Plant Floor Illustration
Imagine a three-step factory: cutting, assembling, and finishing. Finishing can handle half the units that cutting can. No matter how fast cutting runs, finished goods leave at finishing's rate. Cutting extra units only piles inventory on the floor and ties up cash without raising sales.
The One Practice That Changes Everything
Identify your constraint and subordinate every other decision to protecting and elevating it. Stop measuring individual workstations by utilization and start measuring the system by throughput at the bottleneck.
Inventory as a Symptom, Not a Buffer
When managers optimize locally, excess work-in-progress piles up upstream of the constraint. This inventory feels like safety stock but is actually evidence that the system is producing what it cannot yet finish.
- Throughput
- Rate at which the system generates money through sales, not production alone.
- Inventory
- All money tied up in things the system intends to sell, including work-in-progress.
- Operating Expense
- All money spent converting inventory into throughput.
- Constraint
- The single resource whose capacity limits the system's overall throughput rate.
Goldratt ranks these three measures deliberately: raise throughput first, reduce inventory second, cut operating expense third. Most management traditions invert this order, cutting costs before asking whether throughput is being maximized.
A Service Business Contrast
Consider a consultancy where senior reviewers approve every proposal before it goes to clients. Junior consultants can draft proposals far faster than reviewers can approve them. Hiring more junior consultants only deepens the approval backlog. The constraint is reviewer capacity, and adding non-constraint resources elsewhere makes the problem worse, not better.
| Efficiency Mindset | Throughput Mindset |
|---|---|
| Maximize utilization at every step | Protect and feed the constraint |
| Cost reduction is the primary lever | Revenue generation is the primary lever |
| Local metrics guide decisions | System flow guides decisions |
| Idle time is always waste | Idle time at non-constraints is acceptable |
| More output everywhere is better | More output at the constraint is what counts |
When the Framework Holds and When It Bends
The throughput-first logic is strongest when a single, identifiable bottleneck dominates. It is less immediately applicable when constraints shift rapidly, when demand itself is the binding limit rather than internal capacity, or when the system produces many products with wildly different resource profiles.
- Identify: Find the one step that limits overall system output right now.
- Exploit: Get maximum useful output from that constraint without major investment.
- Subordinate: Align all other steps and policies to serve the constraint's pace.
- Elevate: If throughput is still insufficient, invest to expand the constraint's capacity.
- Repeat: Once the constraint moves, return to step one and find the new limiting step.
Why the Constraint Always Migrates
Elevating a constraint successfully eliminates it, but never eliminates constraints from the system. Capacity freed at one point shifts the bottleneck elsewhere. Organizations that celebrate fixing the original constraint and then return to local-efficiency thinking lose the gains quickly because the new constraint goes unmanaged.
This migration is not a failure of the method; it is the method working. Continuous improvement under throughput thinking means continuously relocating and managing the constraint, not arriving at a constraint-free system.
Second-Order Effects on Culture and Measurement
Shifting from efficiency to throughput as the primary metric reshapes accountability. Workers and managers near non-constraint steps must accept that idle time is sometimes correct behavior, which conflicts directly with cultures that treat visible busyness as the definition of contribution.
| Domain | Efficiency-First Consequence | Throughput-First Correction |
|---|---|---|
| Performance review | Workers rewarded for keeping machines busy | Workers rewarded for protecting flow to constraint |
| Capital allocation | Invest wherever utilization is lowest | Invest wherever constraint capacity is tightest |
| Scheduling | Fill every step's calendar | Schedule backward from constraint's capacity |
| Cost accounting | Allocate overhead per unit produced | Measure impact on throughput, inventory, and operating expense |
| Improvement projects | Spread across all steps equally | Concentrate on constraint and its immediate inputs |
Objections Worth Taking Seriously
A common objection is that ignoring non-constraint efficiency wastes money. The reply is that spending on non-constraint activity that does not raise throughput is already waste; the question is only whether the waste is visible or hidden inside utilization numbers.
A second objection holds that constraints are hard to identify in complex, multi-product environments. This is fair. In highly variable or project-based systems, the constraint may be a policy, a approval sequence, or a skill rather than a machine, but it still exists and still governs output.
The search is harder, not unnecessary.
Throughput is a system property. Efficiency is a local illusion.
More structured editions
Reading something else? Faroa builds a structured edition of any book you name. Start free