Find the bottleneck: why the whole line runs at the speed of its slowest step

1. Before you start

A bottleneck is the single slowest step in a process — the one that sets how fast the whole thing can finish, no matter how quick every other step is. A tiny example: if you wash dishes at 20 plates an hour and your partner dries at 12 an hour, the pair of you clear 12 plates an hour, not 16 and not 20. Wash faster and nothing changes; the drying rack is the bottleneck, and only drying faster moves the number. That single idea — the slowest step governs the total — is the whole course.

Three honest statements before you start:

This is a Decide course. You read a situation, learn the idea, and make a call. You do not write or run any code, and there is no calculator to fill in — the arithmetic here is small enough to do in your head, and doing it in your head is the point.
The company in the case, Halstead Instruments, is a composite — an invented manufacturer built from ordinary, realistic dynamics. No number or claim here describes any real company.
This is not a certification. It proves, to you, that you can locate a constraint and defend where to spend the next hour or dollar to make a line go faster.

If you can compare a few station speeds and say which one holds everyone up, you can do this course.

2. The Situation

Halstead Instruments runs one assembly line that builds a handheld barcode scanner, and orders are piling up faster than the line ships them. The plant manager has a budget to speed exactly one thing up — a faster machine, an extra worker, an overtime shift — and every station supervisor is lobbying to have their own station chosen. Spend it on the wrong station and the line ships not one extra unit, and the money is gone for the quarter.

3. What you’ll be able to do

After this course you will be able to:

Find the bottleneck in a flow process by reading the station rates and naming the single slowest step that sets the whole line’s output — and say why the average or the busiest-looking station is the wrong place to look.
Judge where an improvement actually helps: decide whether an hour, a machine, or a headcount spent at a given station lifts the line’s output or is wasted, and say what makes the answer flip.
Apply the identify-exploit-subordinate-elevate sequence to a real decision, and explain why a station’s local efficiency number can look excellent while the line as a whole runs slow.

4. Prerequisites & time box

Prerequisites: none beyond ordinary business reasoning and comparing small whole numbers. No operations background, no spreadsheet, no code setup — the Decide hall is read-and-decide in the browser. This is an early course in the operations track and assumes no prior Decide course.

Time box: about 22 minutes of reading (measured), plus your own thinking time on the call. That is under the 25-minute cap for a concept course.

Difficulty: 3 / 8 — a new-manager decision: a couple of interacting factors and one real judgment call, where the obvious answer is often the trap.

Free-tier honesty: no signup, no paid tool, requires_gpu: false.

5. The case & where the numbers come from

Halstead Instruments is a composite company: an invented contract manufacturer whose figures are in-course assumptions chosen to make the reasoning clean, not drawn from or claimed about any real firm. The framework used to reason about it — the Theory of Constraints and its five focusing steps, from Eliyahu Goldratt’s 1984 book The Goal — is a standard operations concept, cited in section 11. The general terms (throughput, work-in-process, equipment efficiency) are likewise standard and cited there.

Halstead’s line has five stations in sequence. Each unit passes through all five in order. The only case data is each station’s measured output rate — how many finished scanners’ worth of work that station can do in an hour when it runs. Every figure below is an in-course assumption for the teaching case:

#	Station	What it does	Rate (units/hr)
1	Cut & stamp	Cuts and stamps the metal chassis	50
2	Board assembly	Places and reflow-solders the circuit board	20
3	Housing & seal	Fits the housing and weather-seals it	35
4	Calibrate & test	Calibrates the scanner and runs the test rig	25
5	Pack	Boxes and labels for shipment	60

These five rates are the entire dataset. Everything the course concludes about Halstead is reasoned from this one table.

6. The Concepts

This is a judgment course, so the teaching here carries the framework in full: the call is the content, and getting it right depends on understanding why the rule holds, not just stating it. We build up four ideas in the order you would use them at Halstead’s line — where the ceiling comes from, where an improvement is worth anything, how to work the constraint step by step, and why the efficiency dashboard lies — and apply each to the table above as we go.

The bottleneck sets the throughput

Throughput is the rate at which the whole line produces finished units. In a line where every unit must pass through every station in sequence, the throughput can never be faster than the slowest station — because nothing can leave the line quicker than that step can pass work through it. That slowest station is the bottleneck (or constraint).

Read Halstead’s table. The rates are 50, 20, 35, 25, 60. The slowest is board assembly at 20 units/hr, so the whole line ships 20 units/hr — full stop. Not the average of the five rates (which is (50 + 20 + 35 + 25 + 60) ÷ 5 = 38/hr), and not the pace of the fast stations. Cut & stamp can prepare 50 chassis an hour, but board assembly can only consume 20 of them, so 30 chassis an hour pile up in front of station 2 as work-in-process and the line still ships 20. The bottleneck is a governor on the entire line.

The judgment this forces: the busiest, most crowded-looking station is often not the bottleneck, and the bottleneck is often quiet. A bottleneck station is never starved — work is always waiting for it — so it can look calm and fully fed, while the pile of unfinished chassis actually sits upstream of it. Do not locate the constraint by which station looks frantic; locate it by the rates.

Assumptions and limits. This clean “slowest step = throughput” rule assumes a stable line where every unit flows through every station in the same order, one bottleneck at a time, and rates you can actually measure. Two things break it. First, the constraint can be external: if the market only wants 15 units/hr, then demand — not board assembly — is the binding constraint, and speeding the line up just builds unsold inventory. Second, with a changing product mix the bottleneck can float from station to station, so you must re-check the rates, not assume last month’s constraint is this month’s.

An hour saved where it counts

Here is the rule that turns the idea into decisions: an hour of improvement at a non-bottleneck station buys nothing; an hour at the bottleneck lifts the entire line.

Watch it at Halstead. Suppose an engineer tunes the pack station and raises it from 60 to 80 units/hr. The line’s throughput after that change is still… 20/hr, because board assembly still caps it. The pack station was never the limit, so making it faster changes nothing the customer ever sees — it is a mirage of improvement. The same is true of speeding cut & stamp from 50 to 90: the line still ships 20, and you have simply grown the pile of work-in-process faster.

Now spend that hour at the bottleneck instead. Add a second reflow cell to board assembly and raise it from 20 to 26 units/hr. The line’s throughput rises to 25/hr — because once board assembly can do 26, the next slowest station, calibrate & test at 25, becomes the new ceiling. One improvement at the constraint lifted the whole line by 5 units/hr (20 to 25); the identical effort at pack lifted it by zero. That asymmetry is the practical heart of the Theory of Constraints: not all improvements are equal, and only the ones at the constraint reach the customer.

Assumptions and limits. The lift from elevating the bottleneck is capped by wherever the next constraint sits — here you gained 5/hr, not 6, because calibrate & test took over at 25. So the payoff from a bottleneck improvement is only as large as the gap to the second-slowest step. Beyond that gap you are back to buying nothing until you move the new constraint too.

The five focusing steps

Goldratt packaged this into a repeatable loop of five focusing steps. Run them in order on Halstead’s line:

Identify the constraint. Read the rates and name the slowest step: board assembly, 20/hr.
Exploit the constraint. Get the most out of it before spending a cent — make sure it is never idle for want of parts, never runs scrap, and never wastes its scarce time on units that will later fail. An hour the bottleneck spends on a board that fails test is throughput you already owned and threw away. Exploiting can lift output for free.
Subordinate everything else to the constraint. Pace the non-bottleneck stations to the bottleneck’s rate instead of letting them run flat out. Cut & stamp does not need to run at 50/hr; running it there only builds a 30/hr mountain of work-in-process that ties up cash and hides problems. The fast stations serve the constraint, not their own busyness.
Elevate the constraint. Only now spend money to raise its capacity — the second reflow cell, an added shift on station 2, a faster machine. This is the step the budget in section 2 is for, and it is deliberately fourth, after the free gains of exploiting and subordinating.
Repeat — and beware inertia. Once you elevate board assembly past 25, calibrate & test becomes the new constraint. Go back to step 1. The danger Goldratt names is inertia: continuing to protect and pamper the old bottleneck out of habit after it has stopped being the constraint.

The order matters. A plant that jumps straight to step 4 — buying capacity — before exploiting and subordinating often spends money to fix a constraint it was actively starving with bad scheduling.

Why local efficiency misleads

Most plants measure each station on its local efficiency or utilization — the percentage of time it was busy producing. That metric quietly pushes exactly the wrong behaviour. Rewarded on utilization, the cut & stamp supervisor runs the station at 50/hr all shift to keep it “100% efficient” — and every hour produces 30 units of work-in-process that board assembly cannot consume. The station’s local scorecard looks excellent while it actively harms the line: cash is trapped in piles of half-built units, the floor fills with inventory, and throughput is still 20/hr.

The lesson is that the only efficiency that pays is throughput of finished units, not the busyness of any single station. A non-bottleneck running below 100% is not a problem to be fixed — it is a station correctly subordinated to the constraint. Idle time at a non-bottleneck is free; idle time at the bottleneck is the most expensive thing in the plant. Managing to local efficiency numbers inverts that truth and is why so many busy plants ship so little.

Why this and not “keep every station fully loaded”? Because a fully loaded non-bottleneck does not add a single shippable unit — it converts raw material into work-in-process that waits. Full utilization feels like productivity and shows up green on the dashboard, but the line’s output is set at the constraint no matter how busy everyone else is. The genuine limit of this view is the opposite error: subordinating does not mean letting a non-bottleneck run so lean that a hiccup starves the constraint. You keep a small buffer of work in front of the bottleneck precisely so it never goes hungry — subordinate to protect the constraint, not to strangle it.

7. Your Call

You have found one line’s constraint from a standing start. Now a different plant’s decision lands on your desk — and this one comes with a tempting machine quote attached.

Cedar & Coil is a composite manufacturer of upholstered armchairs. Each chair moves through five stations in sequence, and every figure below is an in-course assumption for this transfer case:

#	Station	Rate (chairs/day)
1	Frame build	40
2	Foam & cushion	55
3	Upholstery (sew & wrap)	20
4	Finish & inspect	30
5	Pack & ship	50

A supplier is pitching a $180,000 automated foam-cutting machine that would raise foam & cushion from 55 to 90 chairs/day, and the operations lead is excited about the number. Separately, orders are running at about 24 chairs/day and the sales team is confident it can sell everything the plant can finish. Your job is to decide where the money should actually go.

How this differs from the taught case — the transfer. This is a different company and sector (upholstered furniture, not electronics assembly) with different figures you must re-read, and it asks a different decision type (choose which capital investment lifts output, rather than diagnose a line from scratch). It also adds a new constraint the Halstead case did not have: a concrete machine quote aimed at the wrong station, plus a stated demand figure that lets an external constraint enter the picture. The core concept is the same: locate the bottleneck, then judge where an improvement actually reaches the customer.

8. Self-check

Before you write the memo, make sure you can say each of these in one line, without an answer key:

State your call in one sentence: where should Cedar & Coil’s money and effort go, and what is the line’s throughput today?
Name the one station that is the current constraint, and the fact — its rate versus the others — that proves it.
Name the single change that would flip your call (for example: if demand jumped well above the line’s capacity, or if the machine quote had been for the upholstery station instead of foam).
Say what buying the foam machine would cost Cedar & Coil in the currency that matters: not extra shipped chairs, but cash trapped in a taller pile of work-in-process.

If any of these is fuzzy, reread section 6 — the slowest-step rule and the five focusing steps are the whole course.

9. Stretch

Push the thinking further on your own:

Suppose the $180,000 could instead be split across a second upholstery cell (+12/day) and a faster finish-and-inspect rig (+8/day). In what order would you spend it, and how far does each dollar lift throughput before the constraint moves again? (Trace the constraint hopping from station to station as you spend.)
The upholstery station scraps one chair in ten at final inspection. How does that change where the real constraint sits, and what would “exploiting the constraint” be worth here before you buy any capacity at all?
This is the genuinely hard one: Cedar & Coil’s product mix is about to shift toward a recliner that needs twice the upholstery time but half the finishing. Does your bottleneck move, and how would you decide where to invest when the constraint floats with the mix rather than sitting still?

10. Ship it — your decision memo

Write a one-page memo to Cedar & Coil’s plant manager. State the call (do not buy the foam machine; upholstery at 20/day is the constraint, so put the money and the free exploit-and-subordinate gains there — and watch demand, because at 24 orders/day the market becomes the limit once the line passes it). Give the reasoning in a few lines: name the bottleneck and its rate, show that faster foam leaves throughput at 20/day while lifting upholstery moves it toward 30, and note where the next constraint waits. Name what you rejected (the foam machine; running every station flat out; chasing line speed past demand) and why. Name the one thing that would change your mind (a machine quote aimed at the actual constraint, or demand climbing well above capacity). Keep it to a single page a manager grasps in two minutes. This memo is your own argued claim — not a credential.

11. Sources

Halstead Instruments and Cedar & Coil, and every figure attached to them, are composite — constructed from ordinary, realistic manufacturing dynamics for clean teaching, not drawn from or claimed about any real company. The framework and terms used to reason about them are standard operations concepts; references below.

Concept / claim	Source (publisher)	URL	Accessed
Theory of Constraints and the five focusing steps (identify, exploit, subordinate, elevate, repeat)	Wikipedia — Theory of constraints	https://en.wikipedia.org/wiki/Theory_of_constraints	2026-07-20
Eliyahu Goldratt and The Goal (1984), the origin of the Theory of Constraints	Wikipedia — Eliyahu M. Goldratt	https://en.wikipedia.org/wiki/Eliyahu_M._Goldratt	2026-07-20
Throughput as the rate a system produces finished output	Wikipedia — Throughput	https://en.wikipedia.org/wiki/Throughput	2026-07-20
Work-in-process inventory that accumulates ahead of a constraint	Wikipedia — Work in process	https://en.wikipedia.org/wiki/Work_in_process	2026-07-20
Local efficiency / utilization measures and their limits	Wikipedia — Overall equipment effectiveness	https://en.wikipedia.org/wiki/Overall_equipment_effectiveness	2026-07-20

Next up

Finished this call? Continue the Operations & Process track:

Triage the queue: what to work next when the backlog is winning → · Browse all courses