A Practical Guide

Unlocking plant capacity before you spend.

How to find the binding constraint and apply the Five Focusing Steps — identify, exploit, subordinate, elevate, return — before a capital request goes anywhere near the plan.

Most plants already have the capacity a growth request is asking for. It is sitting inside the operation, unmeasured, because the dashboard was built to show utilization rather than throughput. This guide walks through how to find that capacity, and how to apply the Five Focusing Steps before a capital request goes anywhere near the plan.

Why this matters before you spend a dollar

A plant asked to produce more usually reaches for the same two levers: run more hours, or buy more equipment. Both work. Both are also the expensive way to get there if the real limit was never capacity in the first place, but the way that capacity was being used.

Every plant has a binding constraint — one resource, one station, one operation that sets the ceiling on how much the whole line can produce. Improving anything else feels productive and changes nothing that ships. That is the trap. A capital request built on downtime data or general improvement activity is a weaker case than one built on constraint evidence, because it cannot say with any precision what the money buys.

The Five Focusing Steps exist to put that precision in place before capital is on the table.

Step 01 · Identify

Find the actual constraint, not the loudest one

The constraint is rarely the station operators complain about most. Complaints track visible pain — jams, alarms, rework — and the constraint is often quiet, simply running at its ceiling while everything around it has slack.

A practical way to find it:

This is dynamic bottleneck identification in practice — the constraint moves with product mix, staffing, and condition, so treat identification as something you repeat, not something you conclude once and file away.

Step 02 · Exploit

Get everything out of the constraint before you touch a budget

Once the constraint is identified, the next move is not a purchase order. It is squeezing every unit of output the constraint can already give you, using variance tree analysis to separate the three loss categories in the right order: downtime, slow running, waste.

In practice this usually means:

Exploitation is nearly always free or close to it. It is also the step most plants skip, because it does not produce a line item that looks like progress on a capital plan.

Step 03 · Subordinate

Make the rest of the line serve the constraint

Subordination is the step with the worst reputation, because it asks other stations to run at less than their own local efficiency on purpose. A station upstream of the constraint that could run faster is told to slow down, or to build a small buffer and then stop, rather than push work toward a station that cannot absorb it.

This is where utilization as a metric actively works against the plant. Utilization measures whether an asset is moving, not whether it is moving the right thing at the right time. A non-constraint station running flat out looks good on a scorecard and produces work-in-process that sits in front of the constraint, doing nothing for cases shipped.

Subordination usually means:

Plant managers whose reviews reward station-by-station efficiency will find this step is the one that gets political first. It is worth having that conversation before the capital request, not after.

Step 04 · Elevate

The point where capital finally belongs

Elevation is buying, building, or adding capacity at the constraint. It comes fourth for a reason: exploitation and subordination typically recover meaningful output with no capital at all, and whatever gap remains after those two steps is the real size of the capital ask.

A capital request built this way carries a different kind of evidence than the usual downtime report:

This is usually a stronger case than a general capacity investment, because it answers the question a capital committee actually asks, which is what, specifically, the money changes.

Step 05 · Return

The constraint does not stay put

Elevating a constraint tends to move it. The station that used to set the ceiling now has headroom, and some other operation — often one that looked comfortable a month earlier — becomes the new limit. Returning to Step 1 is not a failure of the first four steps. It is what a production-constrained plant does whenever the constraint changes, and a demand-constrained plant does whenever mix or volume shifts the load.

Treating the Five Focusing Steps as a cycle, rather than a project with an end date, is what keeps a plant from rediscovering the same trap eighteen months later with a different station in the starring role.

Before the capital request goes in

Two questions are worth asking of any growth request before it reaches a capital plan.

Has the constraint actually been identified, or is this request based on a general sense that the plant is busy? Have exploitation and subordination been applied and measured, or is this request skipping straight from identification to elevation because that is the step with a purchase order attached?

If the honest answer to either is no, there is very likely more capacity already inside the plant than the request assumes — unmeasured, already paid for, and cheaper than anything on the capital plan.

This guide draws on Where’s Your Herbie?

A Theory of Constraints book for discrete manufacturing — food and beverage, CPG, and related operations. It goes further into how to walk a plant, find the constraint yourself, and hold the gains once they arrive.