Five Focusing Steps
Pattern: A named solution to a recurring problem.
The operational core of the Theory of Constraints: a five-step cycle that converts a stated goal and an identified constraint into an unending discipline of system improvement.
Also known as: the five steps of focusing, the focusing process.
The Five Focusing Steps turn TOC vocabulary into operating discipline. Before Goldratt named them in 1987, the field had constraint and throughput as concepts and The Goal as its worked example. Afterward it had a five-line loop a plant manager, hospital director, project lead, or sales leader could run without retelling the novel. Drum-Buffer-Rope, Critical Chain, Replenishment Pull, and the Mafia Offer all specialize the same loop for different constraint types.
Context
The pattern anchors Foundations because every later TOC tool specializes it. It presupposes The Goal: without a stated goal, no resource is constraining anything in particular. It also presupposes Constraint: without a working definition of constraint, Step 1 has no anchor. It produces POOGI, the Process of Ongoing Improvement, by looping Step 5 back to Step 1.
Goldratt named the steps after The Goal had spread constraint language beyond its original plant setting. Plants, hospitals, distribution networks, project organizations, and software shops needed a discipline they could run without assuming everyone had read about UniCo. The TOCICO Dictionary canonized the steps as the operational definition of focusing. The TOC Handbook (Cox & Schleier 2010) treats them as the move that turns TOC from a frame into a practice.
The scale can be a CNC machine, a job shop, a hospital A&E, a project portfolio, a software organization, or a national supply chain. The system changes. The loop does not.
Problem
A system exists to produce goal units. The practitioner can usually name several places where output appears limited: a slow machine, a backed-up department, an overworked specialist, an unreliable supplier, a deadline nobody hits. Work directed at any one of them can look plausible in the staff meeting. Most of it is wrong.
Improving a non-constraint does not move the system, and the inherited measurement frame rewards exactly that mistake. Efficiency targets, utilization quotas, and departmental score-cards pull effort toward whichever resource looks busiest or complains loudest. The system needs a loop that identifies the one limit that binds it now, exploits that limit, aligns everything else to it, and elevates it only after alignment is complete and still insufficient.
Forces
• Only one limit sets system output, but local incentives nominate many candidates. The busiest department is not necessarily the constraint.
• The cheap move is rarely the visible move. Exploiting the constraint often means queue discipline, upstream inspection, shift coverage, and maintenance; elevation means money.
• Idle non-constraint capacity feels wasteful. In throughput terms it is protective capacity; at full utilization it becomes excess work the constraint cannot absorb.
• Constraints move. A rule, schedule, or measurement that protected yesterday’s constraint can become today’s inertia.
• Domain language hides the common loop. Production, projects, supply chain, services, and software use different mechanisms, but each specialization still runs the same five steps.
Solution
Run the system through five steps in order. Return to the first when the fifth applies. Refuse to skip a step.
Step 1: Identify the system’s constraint. Determine which element, rule, belief, or market condition is presently limiting goal-unit production. Constraints come in four types: physical (a machine, person, equipment, or space), policy (a rule, procedure, measurement, or incentive), paradigm (a shared belief that makes the policy natural), and market (demand insufficient to challenge internal capacity). The type determines the rest of the sequence. Constraint Type Identification covers the diagnostic signals; Constraint defines the underlying concept. Step 1 ends when the practitioner can point at the constraint and name its type without hedging.
Step 2: Decide how to exploit the constraint. Make the most of the constraint as currently configured, without new investment. For a physical constraint that may mean staffing it through breaks, inspecting upstream, sequencing its queue by throughput per constraint-hour, reducing setup time, and maintaining it aggressively. For a policy constraint it means changing the rule. For a paradigm constraint it means using the Thinking Processes to surface and replace the belief. For a market constraint it means a Mafia Offer, plus the sales discipline to deliver it.
Step 3: Subordinate everything else to that decision. Run every non-constraint at the rate that keeps the constraint fully exploited, not at its own local optimum. The TOCICO Dictionary gives subordination two components: non-constraints must deliver exactly what the constraint needs when it needs it, and they must not process work the constraint will not absorb. The first protects the constraint from starvation. The second prevents the WIP pile-up that creates local efficiency and global lateness. This is the step practitioners most often skip, and its absence produces Improving Non-Constraints and Local Optimization. The standalone Subordination entry expands the move.
Step 4: Elevate the system’s constraint. If exploitation and subordination are complete and the system still cannot produce enough goal units, add capacity: buy, hire, outsource, build, or otherwise expand the constraint. The order matters. Elevating before exploitation and subordination wastes capital on a constraint whose existing capacity was never fully used. Exploitation Without Elevation names the disciplined refusal of that inversion.
Step 5: When the constraint has been broken, return to Step 1. Do not allow inertia to become the system’s constraint. Elevation moves the binding limit somewhere else. The fifth step tells the organization not to keep the rules, schedules, policies, and incentives that suited the previous constraint after the constraint has moved. Many implementations fail here: production capacity is elevated, the market becomes the new constraint, and the plant keeps running the rules that made sense when the line was constrained. The loop is unending by design.
The five steps form a closed cycle. The cycle is the operational definition of the Process of Ongoing Improvement (POOGI).
flowchart TD
A["1 — Identify the constraint"] --> B["2 — Exploit the constraint"]
B --> C["3 — Subordinate everything else"]
C --> D{Sufficient throughput?}
D -->|No| E["4 — Elevate the constraint"]
D -->|Yes| F{Constraint has moved?}
E --> F
F -->|Yes, return| A
F -->|No, hold| C
For non-profit and government systems, where the goal is not making money and the goal unit must be named separately, the Seven Focusing Steps prepend two steps: state the goal of the organization and define global performance measures. The Dictionary names the seven-step variant explicitly. It is the same loop with goal-setting surfaced.
How It Plays Out
The NCX-10 at UniCo. The Goal anchors the pattern in the field’s memory. Alex Rogo’s plant identifies the NCX-10 as the constraint. It exploits the machine by staffing it through breaks, moving inspection upstream, sequencing the queue by throughput per constraint-hour, and reducing setup time. It subordinates by releasing material at the NCX-10’s rate and suspending local-efficiency measures for non-constraints. When that is not enough, the plant elevates by bringing back older machines whose combined capacity covers the gap. When the constraint shifts to the market, the plant returns to Step 1. The problem is no longer production; it is sales and product mix.
Oxfordshire Radcliffe Hospital A&E. The Oxfordshire Radcliffe implementation ran the steps against the four-hour patient-throughput target. Identification named the assessment-to-disposition step in the A&E pathway, with policy elements layered on top. Exploitation reorganized assessment so the available capacity was used fully. Subordination aligned upstream triage and downstream bed allocation to the assessment pace. Elevation was deferred: no added staff until exploitation and subordination were complete. Within roughly two months, patients seen within four hours rose from 50–60% to 95% or above, with no added staff.
A software organization running flow metrics. A platform team identifies its constraint as the single engineer who reviews every database migration. Exploitation filters obvious rejection cases before they reach the engineer and batches migrations into a daily review block. Subordination directs feature branches, deployment requests, and sprint commitments around the review rate. Elevation, if it follows, adds a second reviewer. The discipline is that the team does not hire until the existing reviewer’s exploitation and the team’s subordination have been completed.
Consequences
Benefits:
• A repeatable improvement discipline. The loop is short, named, and remembered; it does not depend on one implementer being in the room.
• A defensible refusal of non-constraint improvement. The practitioner can explain why an efficiency project outside the constraint will not move system performance.
• Cheap moves before expensive moves. Exploitation before elevation turns capital spending from the default into the last resort.
• A guard against stagnation. Step 5 prevents an implementation that worked once from freezing around yesterday’s constraint.
• Cross-domain portability. The loop carries into DBR, Critical Chain, Replenishment Pull, healthcare, services, strategy, and software flow work.
Liabilities:
• Subordination remains the hard step. It asks non-constraints to run below their local optimum, which conflicts with the measurements most organizations still publish.
• Step 1 can be rushed. Practitioners often name the busiest resource and call identification complete; Breaking the Wrong Constraint follows.
• Cost-world measures reward inertia. A factory whose constraint has moved to the market may still be measured on production-line efficiency, pulling it back toward the previous constraint’s policies.
• The loop requires a stated goal. Without a goal-unit anchor, the identification step is malformed. The Seven Focusing Steps variant surfaces that prerequisite.
• Constraint type changes the move. A market or paradigm constraint is not exploited the way a physical constraint is. The structure is universal; the step-level tactics are type-dependent.
Mabin and Balderstone’s 2000 meta-review of more than 80 TOC implementations is the most cited synthesis behind the pattern’s case claim: implementations that complete all five steps in order outperform implementations that skip Step 3 or invert Steps 2 and 4. The honest reading is that complete implementations are rare. The loop is short because it has to be remembered, not because it is easy to hold.
Related Articles
Complements: Throughput Accounting — Throughput Accounting supplies the measurement frame the steps optimize against; the rule of thumb of raising throughput first and reducing inventory and operating expense second is the accounting reading of Step 2 and Step 3.
Depends on: Constraint — The first step identifies the system constraint; the rest of the loop is meaningless until a constraint has been named.
Depends on: The Goal (concept) — The steps are calibrated against the goal: identify, exploit, subordinate, and elevate are all moves to produce more goal units.
Enforces: Exploitation Without Elevation — The exploit-before-elevate rule is the operational discipline the steps impose; the standalone entry names the rule and the conditions under which it binds.
Measures: Throughput — Throughput is the system-level rate the steps move; an improvement that does not raise throughput has not improved the system.
Prevents: Breaking the Wrong Constraint — Step 1's emphasis on identifying the constraint precisely, not the busiest resource, prevents the antipattern of elevating something that only appeared to be constraining.
Prevents: Improving Non-Constraints — The whole sequence is built to prevent improving non-constraints; misallocating effort to non-constraint resources is the failure mode the steps are designed to refuse.
Prevents: Local Optimization — Step 3 prevents local optimization by directing improvement effort at the constraint rather than at whichever non-constraint complains loudest.
Produces: Critical Chain — Critical Chain is the project specialization of the steps: identify the chain, exploit it with shared protection, subordinate non-chain work, elevate when exploitation is insufficient.
Produces: Drum-Buffer-Rope (DBR) — DBR is the production specialization of the steps: the drum exploits the constraint, the rope subordinates release, the buffer protects the exploitation.
Produces: Process of Ongoing Improvement (POOGI) — Returning from Step 5 to Step 1 produces the Process of Ongoing Improvement; the loop is the operational definition of POOGI.
Produces: Subordination — Step 3 is named subordination; the standalone entry expands the move practitioners most commonly skip.
Refines: Constraint Type Identification — Step 1's identification has to name the constraint's type (physical, policy, paradigm, or market) because the type dictates which of the remaining steps applies.
Sources
• Goldratt and Cox, The Goal (1984; revised 1986, 1992) — the originating treatment, dramatized through Alex Rogo’s plant. The 1986 revision carries the subtitle A Process of Ongoing Improvement, and the worked example of the NCX-10 is the canonical illustration of identifying, exploiting, subordinating to, and elevating a physical constraint.
• Goldratt, What Is This Thing Called Theory of Constraints and How Should It Be Implemented? (1990) — formalizes the five-step structure and names the focusing process explicitly. The pairing of focus with the steps was Goldratt’s choice in this volume; the TOCICO Dictionary later canonized it.
• Cox and Schleier, Theory of Constraints Handbook (2010), Ch. 1 (Goldratt, “Introduction to TOC — My Perspective”) — the contemporary authoritative treatment, written by Goldratt himself, in which the steps are presented as the move that turned TOC from a frame into a practice and the moment of their 1987 naming is recorded.
• Cox, TOCICO Dictionary (2nd ed., 2012), entries “five focusing steps,” “seven focusing steps,” “identify,” “exploit,” “subordinate,” and “elevate” — the certification-grade definitions this entry cross-checks against. The Dictionary is distributed through TOCICO membership rather than the trade book channel and doesn’t carry an Open Library record.
• Mabin and Balderstone, The World of the Theory of Constraints (2000) — meta-review of more than 80 documented TOC implementations across production, projects, supply chain, and service settings; the canonical synthesis of the cross-domain case evidence the pattern’s universality claim rests on.
• Dettmer, Goldratt’s Theory of Constraints (1997) — extended treatment of the steps with explicit handling of policy and paradigm constraints and the operational moves each constraint type requires at Steps 2 and 4.