MapKAI

Day 4 of 5

How Strategy Becomes Behaviour

How do measures and controls shape business behaviour?

Estimated time: 25 minutes

Today you will learn to
  • see how measures redirect attention
  • distinguish learning metrics from outcome metrics
  • recognize gaming and double counting
  • use controls to support strategy rather than replace judgment
Start Lesson

Watch the lesson

Subtitles available where provided

Chinese

The Third Spoon of Chili Oil: A Management Control Systems Fable

English

Beyond the KPI Trap

Listen to the podcast

Why Accurate Metrics Sabotage Your Strategy

01

Opening story or business situation

On an old street in the south stood a Sichuan restaurant called Lantern Shadows. It was small, with a string of old lanterns at the door. Every evening, when the first wok sounded in the kitchen, the whole street began to smell fragrant. Peppercorns met hot oil and released their numbing note; dried chilli turned at the wok’s edge and its fragrance slowly deepened. Regulars said Lantern Shadows was not good because it was the hottest restaurant, but because it knew when heat should rise and when it should stop.

The owner, Aunt Lin, cared most about one dish: boiled beef in chilli oil. It looked simple but was difficult. The beef had to be tender, the broth rich, the red oil bright, and the final spoonful of hot oil poured at exactly the right moment. One second early and the fragrance would not rise; one second late and the chillies would turn bitter.

Later, the old street became a tourist street. More restaurants opened and rents rose each year. Aunt Lin hired a capable new manager, Zhou Ming. After studying the accounts for three days, he said, “Lantern Shadows cannot rely only on old masters’ intuition. It needs clear management.”

He made a scorecard. Every dish had to record serving speed, ingredient cost, gross margin, customer rating, and returns. Each Monday, front-of-house and kitchen reviewed it together. Red numbers needed explanation; green numbers earned rewards. At first it worked. Service was faster, waste fell, margins rose, and more guests left reviews. The card on the office wall shone like a newly polished copper mirror.

But the kitchen’s flavour slowly changed.

To improve speed, chefs pre-cooked some dishes halfway. To cut cost, beef was sliced thinner. To reduce returns, spice became gentler and gentler. To earn higher ratings, servers recommended photogenic dishes rather than the labour-intensive dishes that truly kept regulars coming back.

Every number on the card looked good.

But Aunt Lin found that the boiled beef still arrived with red oil, while its aroma had become shorter. Regulars no longer said, “This wok has strength today.” They only nodded politely. Strangest of all, the kitchen grew quiet. Old masters had once listened to the wok, watched the oil, and smelled the peppercorns; now everyone cooked while watching the service-time number on the wall.

There was a young chef named A Yuan. He wanted to create a new dish: fresh fish with green peppercorn and rattan pepper. It did not rely on the traditional heavy oil and chilli, but on the clean numbness of green peppercorn and the tenderness of fresh fish. He tried many times. First the fish broke apart; then the broth was too thin; then the numbness was too sharp; then customers said it did not feel like Sichuan food. On the fifth try it had a bright aftertaste, but serving time was still too long.

Zhou put the dish on the scorecard. The result looked poor: high cost, slow speed, unstable ratings, and no outstanding margin. “We should stop it,” he said. “It is dragging down kitchen efficiency.”

A Yuan did not argue. He took out records of the failures: which fish broke easily, which green peppercorn held its fragrance longer, which heat kept the fish tender, and which customers would accept a less oily Sichuan flavour. “The dish has not succeeded,” he said, “but each failure has taught the kitchen something it did not know before.”

Aunt Lin read them for a long time. Then she asked Zhou, “If this scorecard had judged my first attempt at boiled beef when I was young, I might have stopped making it long ago.”

Zhou said nothing.

The next day, Aunt Lin did not tear down the scorecard. Without numbers, she knew the restaurant would eventually be dragged down by cost, waste, and disorder. A master’s feel mattered, but it could not excuse never doing the arithmetic.

She simply hung a second card beside it.

The first card continued to govern mature dishes: boiled beef, kung pao chicken, twice-cooked pork, and mapo tofu. Their methods were stable, so speed, cost, quality, and customer feedback should be watched. Mature dishes could not say “today is exceptional” every day.

The second card governed only new dishes. It did not ask how much they earned that week. It asked three questions: What did this experiment validate? What did it rule out? What will change next time? A Yuan’s fish could no longer use the dinner-rush stove; it could be tested only in the afternoon, with a small budget each time and a gate before moving to the next step.

Later Zhou faced another issue. The restaurant’s most profitable new dish was Flaming Chilli Chicken. Servers set it on fire at the table; it looked excellent on a phone. Tourists loved it, ratings were high, and margins were high. By the first card, it belonged on the first page of the menu.

But Aunt Lin did not agree.

She pointed to words under the old lanterns: “Let people far from home taste a flavour with roots.”

“We can make dishes tourists enjoy, and we can earn money,” she said. “But if a dish has spectacle without flavour; if it makes the kitchen sacrifice heat for performance; if it turns real Sichuan taste into a short video, then however high its score, it cannot represent Lantern Shadows.”

From then on, three rules could not be bypassed: ingredients could not be stale; heat could not give way only to speed; and signature flavour could not lose its roots merely to please everyone. Anyone seeking an exception had to record the reason, the owner, and when the result would be reviewed.

Years later, restaurants on the old street came and went. Some had thicker menus, brighter decor, or higher ratings. Yet many people still chose Lantern Shadows.

They said its strength was neither blind faith in old masters nor blind faith in forms.

What it truly protected was the third spoonful of red oil.

The first spoon gives a dish colour. The second releases its aroma. The third decides whether the dish has a soul.

Management is the same. Numbers keep people from confusion and controls keep them from drifting, but good control does not cook every dish into the same flavour. It knows what should be stable, what should be tried, and what no high score may cross: the taste that truly matters.

02

The decision problem

03

Core concepts

Performance measurement

see how measures redirect attention

Why it matters

What behaviour does your current KPI reward?

Management control

distinguish learning metrics from outcome metrics

Why it matters

Which early learning signal matters before financial results arrive?

Incentives

recognize gaming and double counting

Why it matters

What behaviour does your current KPI reward?

04

Visual framework

01Begin with a business situation
02Reveal the financial logic
03Connect the idea to a practical framework
04Apply it through reflection and decision questions

Five-step knowledge chain | Turning strategic intent into control without harm

Clarify the decision → choose information that changes it → predict how metrics change behaviour → match control to uncertainty → put value trade-offs into resource decisions.

01State the decision first

Pricing, cost reduction, resource allocation, and exploration need different information, horizons, and tolerances for error. Clarify the strategic aim, decision owner, deadline, and desired action. Control starts with defining the problem, not listing KPIs.

02Choose information that changes a choice

Information is useful when it changes this decision, not merely when it is precise. Separate costs that change with the option from committed costs, and test capacity, customer value, time, and definitions. Decision support and performance evaluation are not interchangeable; an easy-to-count metric that does not improve a choice is false precision.

03Before rewards attach, predict how people will chase the metric

Bonuses, promotion, budgets, and rankings turn a metric into a behaviour rule. Ask whether everyone maximising it could worsen the real outcome. Check controllability, gaming, short-termism, double counting, and important unmeasured effects. Scores should support discussion, never automatically replace judgement.

04Use correction for mature work and learning for exploration

Stable work with known cause and effect suits diagnostic control: set targets, observe variance, correct execution. New technology, customers, and business models suit interactive control: small staged investment, milestones, and repeated tests of key assumptions. Failure becomes valuable learning only when it leaves transferable evidence and changes a continue, expand, pivot, or stop decision.

05Put purpose into gates, budgets, and exceptions

Define non-negotiable minimum gates such as safety, ethics, compliance, and core customer value. A project that fails a gate cannot continue regardless of its score. Above the gate, evidence-based exceptions may be approved, but reason, owner, review date, and result must be recorded.

05

Practical example

Business example

After AI automation, retain judgement about customer value, learning quality, and long-term options instead of chasing efficiency numbers alone.

06

Common misunderstanding

07

Key takeaways

  • see how measures redirect attention
  • distinguish learning metrics from outcome metrics
  • recognize gaming and double counting
  • use controls to support strategy rather than replace judgment

08

Knowledge check

How do measures and controls shape business behaviour?

09

Reference notes

  1. Simons, Robert & David Geiger. Tennessee Controls: The Strategic Ranking Problem. Harvard Business School Case No. 191-083 (1991; revised 2010).

    HBS decision case.

    Supports the Strategic Ranking Index, verifiability of inputs, weights, double counting, and the proper authority of a model.

  2. Davila, Antonio (2005). “The Promise of Management Control Systems for Innovation and Strategic Change.”

    Oxford University Press chapter.

    Supports control as innovation support: make assumptions explicit, create review rhythm, synchronise resource commitments with evidence, and separate early learning measures from later financial measures.

  3. Quinn, Robert E. & Anjan V. Thakor (2018). “Creating a Purpose-Driven Organization.” Harvard Business Review.

    Organisation-purpose article.

    Supports putting purpose into leader behaviour, resource allocation, promotion, and consequences rather than treating it as a cultural slogan.