Theory of constraints: fixing the slowest step

A manufacturing plant has five machines in sequence. Machine 3 is the slowest — it can process 100 units per hour. The other machines can each process 150 units per hour.

The plant manager improves Machine 1 to handle 200 units per hour. Nothing changes. Total output is still 100 units per hour.

Why? Because Machine 3 is the bottleneck. No matter how fast the other machines run, output cannot exceed what Machine 3 can handle.

The pattern: Every system has a constraint — one element that limits the entire system's performance. Improving anything else does not improve the system.

The framework: Eliyahu Goldratt developed the Theory of Constraints (TOC) in his 1984 book "The Goal." It has become one of the most influential ideas in operations management.

The constraint is where work piles up. Look for queues.

Signs of a bottleneck

• Work waiting to be processed
• Delays happening at the same point consistently
• One person or resource that everyone depends on
• A step that takes longer than all others

Common bottlenecks in business

• The founder's inbox (decisions wait for approval)
• One specialist whose skills are scarce
• A manual process that cannot scale
• A system with limited capacity
• A team that is understaffed relative to demand

The constraint changes

Once you fix one bottleneck, a new one appears. The system's throughput increases until the next slowest element becomes the new constraint.

This is not a problem — it is how improvement works. You keep finding and addressing the next bottleneck.

Goldratt outlined a systematic approach to managing constraints.

Step 1: Identify the constraint

Find the bottleneck. Where is work piling up? What is everyone waiting for?

Step 2: Exploit the constraint

Make the constraint as productive as possible with existing resources:

• Remove any waste in the constraint's process
• Ensure the constraint never sits idle
• Only give the constraint the highest-value work

Do not add resources yet — first make sure the current capacity is fully used.

Step 3: Subordinate everything else

Align all other operations to serve the constraint:

• Do not produce faster than the constraint can handle (you just create inventory)
• Protect the constraint from interruptions
• Prioritize what helps the constraint move faster

Everything that does not help the constraint is secondary.

Step 4: Elevate the constraint

If the constraint is still limiting after exploitation and subordination, add capacity:

• Hire more people for that role
• Buy more equipment
• Outsource that function
• Find alternatives

This costs money, so exhaust free options first.

Step 5: Repeat

Once the constraint is resolved, a new constraint appears. Return to Step 1.

Drum-Buffer-Rope (DBR) is a scheduling method based on TOC. It synchronizes the entire system to the constraint's pace.

The Drum

The constraint sets the rhythm. If the bottleneck can handle 100 units per hour, the whole system should produce 100 units per hour.

Making upstream processes faster than 100 units per hour just creates piles of work-in-progress that the constraint cannot process.

The Buffer

Place a strategic inventory of work in front of the constraint. This ensures the constraint always has something to work on, even if upstream processes have hiccups.

The buffer protects the constraint from variability — late deliveries, unexpected problems, or uneven flow.

The Rope

Control when work is released into the system. Do not start new work until the constraint has capacity.

The "rope" is the signal that connects the constraint to the beginning of the process. It prevents overproduction and keeps work-in-progress under control.

Traditional management optimizes for efficiency — keeping everyone busy. TOC optimizes for throughput — the rate at which the system generates results.

The problem with efficiency

If every department tries to be 100% busy, each will produce as fast as possible. But if downstream processes cannot handle the volume, you get:

• Inventory piling up between steps
• Cash tied up in work-in-progress
• Chaos when the constraint gets overwhelmed

Being busy is not the same as being productive.

Throughput thinking

Only one thing matters: the rate at which finished products or services reach customers and generate revenue.

A "less efficient" system that produces more finished output is better than an "efficient" system clogged with work-in-progress.

Measure how much gets completed, not how busy people appear.

TOC applies beyond manufacturing to any system.

Sales constraints

If the sales team cannot close faster, adding more leads just creates a bigger backlog. The constraint might be:

• Sales capacity (need more salespeople)
• Sales skill (need better training)
• Product-market fit (leads are not qualified)

Marketing constraints

If marketing generates more leads than sales can handle, marketing is not the constraint. Spending more on marketing just wastes money.

Service delivery constraints

If delivery cannot keep up with sales, customers wait. Fix delivery before accelerating sales.

The founder as constraint

In many small businesses, the founder is the bottleneck. Every decision, every client interaction, every strategic choice flows through one person.

This is the most dangerous constraint because it limits the entire business to one person's capacity.

TOC is powerful but not universal.

Complex systems with multiple constraints

Some systems have multiple interacting bottlenecks that shift depending on the situation. The simple five-step model may need adaptation.

Creative and knowledge work

In highly variable work, the "constraint" may not be a stable point. The bottleneck might shift based on the type of project or client.

Over-focusing on one constraint

If you only optimize the constraint and ignore everything else for too long, other parts of the system can deteriorate. Balance is needed.

Political resistance

Telling departments they are "not the constraint" can create conflict. People want to feel important, and TOC can seem to devalue their work.