In het Nederlands
Interactive presentation · click anything for an explanation
Theory of Constraints
The weakest link
first
Every organisation has, at any moment, exactly one brake on its throughput. Improve anywhere else and nothing changes. This session is about finding that brake, and what you do with it next.
Speaker notes
This is a session on the Theory of Constraints, Eliyahu Goldratt's theory from his book The Goal. Not as theory to know, but as a lens to look at your own chain: where is the brake, and are we working on it or are we busy somewhere else.
Three words to hold on to: find, focus, repeat. Find the brake, focus everything on it, and start again as soon as it moves.
This deck rests on the vault note theory-of-constraints and on the way of working of the iAi model.
its weakest link. Every system with a goal is limited, at any moment, by exactly one bottleneck: the constraint. Working harder anywhere that is not the constraint gains nothing for the whole. So the art is not to work harder everywhere, but to find the constraint and focus everything on it.
Speaker notes
If you remember one sentence: a chain is as strong as its weakest link. It sounds like a proverb, but the consequence is hard. Everything you improve in a place that is not the weakest link changes nothing about what the chain delivers as a whole.
Goldratt calls that link the constraint. There is always one. Never zero, and rarely more than one at a time.
Five steps from order to delivery. What does the chain deliver?
Speaker notes
A simplified chain: orders in, planning, production, inspection, shipping. Each step can handle a number of orders per day. Production can do six. So what does the chain deliver? Six. Not the average, not the sum: the lowest.
Click "faster planning". Planning can now do sixteen a day, and throughput stays at six. The only thing that grows is the pile of work waiting for production. That is the yellow stack. Click "faster inspection": same story.
Then click "faster production". Now throughput goes up, to nine. Click once more: production can do twelve, and the brake moves to shipping. That is the whole story in one picture: only improving the constraint counts, and afterwards it moves.
Goldratt measures an organisation by three things
Throughput
The rate at which the system makes money through sales and delivery. Not what is made, but what is delivered and paid for.
Inventory
All the money tied up in the system: material, semi-finished goods, work in progress, but also unfinished projects and open files.
Operating expense
The money you spend to turn inventory into throughput: people, machines, energy, licences.
The winning move is almost always raising throughput, not cutting cost. Cost has a floor, throughput does not.
Speaker notes
Before you look for the constraint, you need to know what you steer on. Goldratt uses three measures. Throughput: what you earn, per day, per week. Inventory: everything that is tied up and has not yet delivered anything. Operating expense: what it costs to keep the system running.
Most improvement programmes sit on the third: cut cost. But cost has a floor. Throughput does not. That is why ToC looks first at what limits throughput.
Improve the system, not one part
Making your product or service: the steps from demand to delivered and paid.
Demand, capacity, supply, cash, attention. Not enough of any one of them: that is the constraint.
Which rules steer work and resources? Sometimes a rule itself is the brake.
Lag: are we reaching the goal. Lead: why not. Compliance: are we following the rules.
"The aim of all science is to substitute visible complexity with invisible simplicity."Jean Baptiste Perrin, physicist (1870-1942). Every organisation looks complex; the constraint is the simplicity underneath.
Speaker notes
Before you improve anything: what are you improving? A system. Every system has five parts. A goal. Work to reach that goal. Resources to do the work. Rules to steer work and resources. And metrics that give feedback: lag, are we reaching the goal; lead, why not; compliance, are we following the rules.
The classic approach improves each part on its own. That gives better parts, not a better system. The sum of local improvements is not a system improvement. The resource you do not have enough of, that is the constraint, and that is where the leverage is. Perrin said it: all science replaces visible complexity with invisible simplicity. In an organisation the constraint is that simplicity.
Found it? Then focus everything on that one link
Which link limits throughput towards the goal? Look where the work piles up.
Get everything out of it without investing: no idle time, no waste, only the right work through it.
Align everything else to the pace of the constraint. Do not run faster than it can handle.
Only once 2 and 3 are exhausted: invest to lift the capacity of the constraint.
The constraint is solved and moves. Back to step 1. Do not let habit become the new brake.
Speaker notes
Goldratt gives five steps. One: identify. Two: exploit, get out what is in it without spending money. A machine that stands idle an hour a day because the operator takes a break is a free hour of capacity. Three: subordinate, the rest of the organisation follows the pace of the constraint. This is the hardest step, because it means other departments deliberately do less.
Four: elevate, now invest. And five: repeat, because the constraint is now somewhere else. The trap is that the rules you made in steps 2 and 3 linger while the constraint has long moved. Then the habit is the new constraint.
The constraint sets the beat, the rest follows
Speaker notes
In production, subordinating is called drum-buffer-rope. The constraint is the drum: it sets the beat. In front of the drum lies a buffer, a small stock of work, so it never stands idle. An hour lost on the constraint is an hour lost for the whole chain; you never make it up.
And the rope ties the inflow to the beat of the drum: no more work enters the system than the constraint can handle. That keeps the pile small and the lead time short. Outside production it works exactly the same: a helpdesk, an analysis team, an approval step.
One to one: from goal to the one experiment
How do you tell, out of everything you can improve, the one thing you must change? By pairing one "what" with one "how" at every step. Never one to many.
One goal, income or impact. One resource you do not have enough of. Quick check: what is the work waiting for?
The shortage is made of problems. One of them gives at least half the gap. That is the problem for now.
The solution collides with the status quo. Each side has its own pros and cons. Write the conflict out.
A solution without trade-offs: all the pros, none of the cons. That is the requirement for the solution.
Test one assumption: do we now have more of the scarce resource? Then back to pairing 1.
Speaker notes
The five focusing steps say where to push. Alan Barnard, a student of Goldratt, turns that into a method you can repeat at every level: five pairings, each a one-to-one relationship between a what and a how. Goal and constraint: one goal, one resource you do not have enough of. Constraint and problem: the shortage is made of problems, and Pareto is fractal, so one problem gives at least half the gap.
Problem and conflict: the solution is one side of a conflict with the status quo, and until you write that conflict out you get stuck or you overreact. Conflict and innovation: a solution without trade-offs, all the pros and none of the cons, that is the requirement. And innovation and experiment: test one assumption, do we now have more of the scarce resource. Then back to the start: same constraint, next problem; constraint moved, start again.
- Busy is not the same as constraintBeing busy can also mean doing work the constraint will never process. Look where the work piles up, not where the sweat is.
- Local efficiency can be harmfulA department running faster than the constraint builds inventory, hides the real problem and costs money.
- But: ToC is not everythingToC says where to push, not how hard and not whether you may. Limits on people, market and mandate still stand. And sometimes the constraint is a rule, not a machine.
Speaker notes
The counter-argument we hear most often: with us it is everywhere. It feels that way, but it is rarely true. Being busy is not the same as being the constraint. Look at where the work is waiting, not at who runs hardest.
And an honest footnote: ToC tells you where to push. Not how hard, and not whether you may. People, market and mandate have limits that are not in the model. Sometimes the constraint is not a machine or a department but a rule. That is what the next slide is about.
- "Why are you asking about leads?"That is not the problem you are solving. You are looking for more capacity, at the lowest cost and the lowest risk. Anything that does not add to that is distraction, however interesting.
- Audits and best practices make the list longCompare every department with best practice and you find a gap everywhere. Every gap goes on a list, and management is overwhelmed.
- One to one is the only way outOne goal, one constraint, one problem, one conflict, one innovation, one experiment. Not one to many. That is what holds the focus when everything looks interesting.
Speaker notes
An example of discipline. In a workshop by Alex Hormozi everyone had first named their goal and their constraint. A chiropractor short of capacity then asked about lead generation through Facebook. The answer was blunt: why are you asking about that? That is not your problem. You are looking for more capacity, as cheaply and safely as possible.
That is also why audits and best-practice comparisons go wrong so often. You find a gap everywhere, every gap goes on the list, and nobody gets to steer any more. The only way out is to guard the one-to-one relationship: one goal, one constraint, one problem. The rest can wait, even when it is interesting.
Four places where the constraint hides
A step that cannot keep up
A machine, a team, one person with a skill nobody else has.
A rule that slows throughput
An approval step, a batch size, a "that is how we do it here". Often clever once, now a brake.
Too little demand
The chain can do more than is sold. Then the constraint sits outside the door and sales is the link.
Knowing sits in one head
The work can only be done by whoever knows it. Then handover is the constraint, not execution.
Speaker notes
Four places to look. The first is the familiar one: a step that cannot keep up. The second is missed most often: policy. A rule that was once clever, an approval that holds everyone up, a batch size from another era. Policy constraints are cheap to solve and hard to see.
The third is the market: if the chain can do more than is sold, sales is the link. And the fourth is knowledge: if the work can only be done by whoever knows it, handover is the brake. You recognise it by "you will have to ask her".
Which process do we work out first? The constraint.
- CoreWhat the organisation exists for, apart from its processes.
- Main groups and chain processesFrom demand to delivery, in big steps.
- First high-level SIPOCPer chain process: what comes in, what goes out.
- The choice: which detail process firstThis is where ToC fits. The process that limits the client's throughput most is the one we work out first, down to requirements. Not the easiest, not the most fun.
- Requirements, realisation, operationsFrom need to technology, and into use.
- Why hereYou cannot work everything out in depth at once. The order decides where the first value lands. ToC makes that choice traceable instead of a gut feeling.
- What it looked like at a manufacturerCore and chain processes mapped. Order processing, turning orders into a workable production schedule, pushed delivery reliability below the strategic 95 percent. That process first, the ones around it later.
- And at homeThe same question is on every management meeting: what is our one constraint right now, and are we putting our hours there?
Speaker notes
For us, ToC is not a loose theory but the engine under the iAi model. After we have taken an organisation apart, core, main groups, chain processes, a first SIPOC, we face the sharpest choice of the whole project: which detail process do we work out first, down to requirements.
ToC gives the answer: the process that limits the client's throughput most. At a manufacturer that was order processing, because it pushed delivery reliability below 95 percent. That process first. And we ask ourselves the same question in every management meeting.
Find your constraint in four rounds
- Draw the chainFrom demand to delivered and paid, in five to eight steps. One line, no org chart. Twenty minutes.
- Where does the work pile up?Queues, full inboxes, "it is with", open tickets. The pile lies in front of the constraint. Put a red dot. Fifteen minutes.
- What does the constraint do that it does not have to?Exploit without investing: waiting, searching, rework, meetings. Cross it out. Fifteen minutes.
- Who runs faster than the constraint?Subordinate: what would that department do differently tomorrow? And which rule holds the constraint back? Twenty minutes.
On the board
One line with the chain steps, a number per step (orders, files, tickets per week), one red dot.
Rule of the game: one constraint. When in doubt you choose, and you test it in the next round.
The result
One sentence: "Our constraint is X, because Y piles up there, and next week we do Z to exploit it."
That is the detail process we work out first.
Speaker notes
Now you. Four rounds, about seventy minutes in all. Round one: draw the chain on one line, five to eight steps, from demand to delivered and paid. No org chart, a line.
Round two: where does the work pile up? That is the clue. The pile always lies in front of the constraint. Put a red dot. Round three: what does that step do that it does not have to? That is exploiting. Round four: who runs faster than the constraint, and which rule holds it back? That is subordinating and policy.
The result is one sentence. Our constraint is X, because Y piles up there, and next week we do Z. That sentence is the start of the detail process we work out first.
Fill in your five pairings
Type straight into the boxes. The text stays saved in this browser, and the button copies everything as text for the report.
Speaker notes
This is the action plan after the four rounds. The red dot from round two is pairing 1: goal and constraint. Fill in the rest together, out loud. Pairing 2: which one problem explains most of the shortage? Pairing 3: what do we gain and lose if we solve it, and if we do nothing? Pairing 4: what does a solution without trade-offs look like? Pairing 5: what do we test next week, and how do we see that it works?
The text stays in the browser and goes to the report as text with the button. One line per pairing is enough. What does not fit in one line is not a choice yet.
Find, focus, repeat
Three images
- FindThe chain in your hands: you are not looking for the strongest link, but for the one that breaks first.
- FocusThe drum: one instrument sets the beat, the rest of the band does not play faster than the drummer.
- RepeatThe repaired link: as soon as it is strong, another one is the weakest. Back to the chain.
Three moments to recall it
- Day 1Tell a colleague in three sentences what your constraint is and why.
- Day 3Name the five focusing steps from memory. Which two cost no money?
- Day 10Look at what you improved this week. Was it on the constraint, or next to it?
Speaker notes
Three images to hold on to. The chain: you look for the link that breaks first. The drum: one instrument sets the pace, the rest does not play faster. And the repaired link: as soon as it is strong, another one is the weakest.
And recall it instead of rereading it. Tomorrow retell it, in three days the five steps from memory, in ten days the honest question: was what I improved this week on the constraint?
The rest waits.
Our advice: do not start an improvement project, a system setup or an AI application without first writing down the one sentence: our constraint is X, because Y piles up there. Work out that process first. The rest follows, and often cheaper.
Speaker notes
Back to the start: improve the weakest link, the rest waits. Find, focus, repeat.
Our advice to take home: do not start an improvement project, a setup or an AI application without first writing down that one sentence. Our constraint is X, because Y piles up there. That process first. If you want to do that together, a conversation is the next step.