Theory of Constraints
The Theory of Constraints (TOC) is a management and systems methodology developed by Eliyahu Goldratt, first presented in his 1984 novel The Goal. The central claim is simple: every system has exactly one constraint that limits its throughput at any given time, and improving anything other than that constraint is waste.
Core Idea
A system’s output is determined by its bottleneck. A factory that can machine 100 parts per hour but only assemble 60 is a 60-part-per-hour factory regardless of how fast machining gets. Investing in faster machining before fixing assembly is misallocation — it produces work-in-progress inventory, not output.
This is not a metaphor. It is a direct consequence of flow dynamics: in any sequential or networked process, the slowest stage determines the rate of the whole. Optimising non-bottleneck stages increases local efficiency while leaving system throughput unchanged. Worse, it can increase costs (more inventory, more coordination overhead) without corresponding revenue.
The Five Focusing Steps
Goldratt formalised the approach as an iterative cycle:
- Identify the constraint — find the stage that currently limits system throughput
- Exploit the constraint — maximise the output of the bottleneck with existing resources (no new investment, just better utilisation)
- Subordinate everything else to the constraint — align all other processes to serve the bottleneck rather than optimising locally
- Elevate the constraint — invest in expanding the bottleneck’s capacity (new equipment, more staff, redesigned process)
- Repeat — once the constraint is broken, a new constraint emerges elsewhere in the system. Return to step 1.
The critical insight is step 5: constraints migrate. Solving one bottleneck doesn’t eliminate constraints from the system — it moves the binding constraint to a different stage. The system always has exactly one constraint; the question is where.
Constraint Migration
This is where TOC becomes more than an operations management tool. The pattern — solve a constraint, expose the next one, solve that, expose the next — appears across domains:
- Manufacturing: machine speed → assembly capacity → raw material supply → distribution logistics
- Software: compute → memory → I/O → network bandwidth
- Organisations: hiring → onboarding → coordination → decision-making speed
The constraint doesn’t just move randomly. Solving a constraint at one stage changes the relative costs and capacities at adjacent stages, which determines where the next bottleneck appears. This is why local optimisation fails — it doesn’t account for how improvements at one stage reprice the demands on neighbouring stages.
Throughput Accounting
Goldratt rejected traditional cost accounting as misleading for operational decisions. He proposed three metrics:
- Throughput (T): the rate at which the system generates money through sales — not production, sales. Producing unsold inventory is not throughput.
- Inventory (I): money tied up in things the system intends to sell — raw materials, work-in-progress, finished goods.
- Operating Expense (OE): money spent turning inventory into throughput — labour, overhead, utilities.
The goal is to increase T while decreasing I and OE. Traditional cost accounting optimises OE locally (make each department cheaper), which often increases I (more inventory piling up at bottlenecks) and leaves T unchanged. Throughput accounting keeps attention on the constraint.
Drum-Buffer-Rope
Goldratt’s scheduling method for production systems:
- Drum: the constraint sets the pace (the “drumbeat”) for the entire system
- Buffer: a time buffer protects the constraint from upstream disruptions — ensure the bottleneck is never starved for work
- Rope: a communication mechanism that ties the release of new work to the constraint’s pace — prevents overproduction upstream
The principle is that the constraint should never be idle (it’s the most expensive lost time in the system) and non-constraint stages should produce only what the constraint can absorb.
Thinking Processes
Later work extended TOC beyond manufacturing into general problem-solving through a set of logic-based tools:
- Current Reality Tree: maps cause-and-effect to identify the core constraint
- Evaporating Cloud: surfaces the assumptions behind apparent dilemmas
- Future Reality Tree: tests whether a proposed solution actually resolves the constraint
- Prerequisite Tree: identifies obstacles to implementation
- Transition Tree: sequences the actions needed
These are essentially structured methods for ensuring you’re solving the right problem before optimising.
Relation to Other Frameworks
TOC shares DNA with lean manufacturing (Toyota Production System) but differs in emphasis. Lean targets waste elimination across the entire process. TOC argues that waste elimination at non-constraints is irrelevant to throughput — only the constraint matters. In practice, organisations often use both: TOC to identify where to focus, lean to optimise once the focus is set.
The constraint migration pattern also appears in economies of scale analysis — scaling resolves one cost constraint but eventually exposes diseconomies (coordination overhead, communication degradation) that become the new binding constraint.