Hello TOC

Advanced Fundamentals · Lesson 3 of 8

Thinking like a scientist

Why the conventional company loses its direction, why data alone doesn’t settle arguments, and how predicted effects let you validate a cause you can’t see.

14 min · Not completed yet

The directionless company

How does a company try to make money? It wants to increase sales and decrease costs. So functions pull in different directions: marketing wants more variety and short lead times; production wants long runs of few products; purchasing wants cheaper raw material; finance wants less inventory; HR wants fewer people; quality wants better quality. Each pull is sensible on its own.

The conventional way of managing:

  1. Decide a vision and derive a mission.
  2. Prepare an annual business plan with goals and objectives.
  3. Split the goals among departments.
  4. Departments set annual targets and pass them down to their functions.
  5. Each function launches improvement programmes to hit its targets.

The hidden assumption: the sum of improvements in each part equals the improvement of the whole.

What happens in practice? Improvement projects from different functions conflict. Departmental priorities override company priorities. Each person starts work in many areas, so focus on the real leverage point is lost. Reviews eat the time, while work-in-progress piles up with no net output. People are judged on targets and dates rather than on completed actions. An action that helps one element hurts another: cut inventory, and materials run short, so lead times grow; push for on-time delivery, and quality or cost suffers. Management walks a tightrope between sales and cost, with no way to know the right balance, and the strategy gets lost along the way. To senior management, the business looks hopelessly complex, because they can’t see the cause-and-effect beneath it.

What makes something scientific?

The philosopher Karl Popper offered a simple test: a theory is scientific only if it is falsifiable, only if some observation could prove it wrong.

The problem with data

We’re often told to “let the data speak.” But in many debates, both sides present data, carefully chosen, and each set seems to prove its point. Data doesn’t speak for itself; the speaker and their biases do. People look for data that confirms what they already believe. This is confirmation bias, and it is everywhere in management discussions.

Pause and think: two managers argue about why deliveries are late. Each brings a spreadsheet that supports their view. How can the argument be settled?

  • By asking what else each explanation predicts, and checking whether those other effects exist
  • By collecting more spreadsheets
  • By letting the senior person decide
  • By averaging the two views

Each explanation, if true, must produce other effects too. Find the effects one explanation predicts and the other doesn’t, and check them.

Predicted effects

Physicists faced this problem when they couldn’t observe causes directly. Their answer: in an interconnected system, a cause always has several effects, spread out in time and place. If one of the predicted effects is missing, the existence of the cause is in doubt, and the theory must change. So we can validate a cause even when we can’t see it, by checking the effects it predicts.

A doctor works exactly this way:

Elimination: high fever fits viral fever, typhoid and dengue. An enlarged liver rules out viral fever. Rashes and bleeding gums leave only dengue. Validation: if it is dengue, we also expect joint ache and loss of appetite. If they are there, our confidence grows; if not, we think again.

This is the logic behind TOC’s cause-and-effect tools. It lets a team reach agreement on what is really going on, without relying on whose data is louder.

Check your understanding

Answer all five. Get four right to complete the lesson.

1. The hidden assumption of conventional management is that…

  • People resist change
  • Customers are always right
  • The sum of improvements in each part equals the improvement of the whole
  • Data never lies

In a system of dependent parts, it doesn’t hold.

2. According to Popper, a theory is scientific only if…

  • Most experts agree with it
  • It is falsifiable: some observation could prove it wrong
  • It has lots of data
  • It is complicated

A claim nothing could disprove isn’t science.

3. Confirmation bias means people…

  • Confirm every order
  • Avoid data
  • Always change their minds
  • Look for data that supports what they already believe

So data alone rarely settles a debate.

4. In predicted-effect thinking, if an effect the cause should produce is missing…

  • The existence of that cause is in doubt
  • The cause is confirmed
  • It doesn’t matter
  • More data is needed

Every real cause has several effects.

5. In the diagnosis example, which observation rules out viral fever?

  • High fever
  • Joint ache
  • An enlarged liver
  • Loss of appetite

Viral fever wouldn’t predict it.