Define Before You Solve

The most common mistake in problem-solving is not getting the answer wrong. It is answering the wrong question.

In quality management, there is a discipline called DMAIC — Define, Measure, Analyse, Improve, Control. It was developed for manufacturing, where it eliminated defects on production lines. The first phase is Define: establish what the actual problem is before you touch anything.

It sounds obvious. It is almost never done.

Modena, Italy — 185,000 people, in the Po Valley — has a traffic problem that everyone can see. Car mode share sits at 63%, according to the city's own sustainable mobility plan. There are 160,000 registered motor vehicles, roughly one per resident including infants. Commuters spend over 200 hours per year driving and around €5,700 on car ownership. A fit cyclist completes a typical urban trip in 36% less time than a car.

The instinct is to fix the traffic. Build a bypass. Widen the road. Add capacity.

But transport economists have known since 1962 that adding road capacity generates new demand. Anthony Downs described the phenomenon in Traffic Quarterly that year. Duranton and Turner confirmed it empirically in the American Economic Review in 2011: new highway capacity fills within one to five years. The solution reproduces the problem.

This is what happens when you skip the Define phase. You see congestion and reach for the road. You see the symptom and treat it. You solve what is visible rather than what is causal.

When you actually define the problem — when you follow the data instead of the complaint — you find that the real cost is not the traffic itself but the system that produces it. Modena has 120,000 parking spaces. The implicit annual subsidy for that free parking is approximately €41.8 million, calculated using the methodology in Donald Shoup's The High Cost of Free Parking. Every new housing unit is required by law to include 1.0–1.5 parking spaces, adding €15,000–25,000 to every apartment's price. The city builds car dependency into its physical structure, then wonders why 63% of trips happen by car.

Stockholm understood this. When it introduced congestion pricing in 2006, traffic in the metropolitan area dropped 20–22%, according to Jonas Eliasson's research in Transportation Research Part A. CO₂ emissions fell 2–3%. Not because they built more roads. Because they redefined the problem: not insufficient capacity, but underpriced demand.

Meanwhile, nearly one-third of Modena's residents — 54,500 people — cannot drive at all, due to age, disability, economics, or legal status. They depend entirely on a system designed around a mode they cannot use. The bus rider sits in the same traffic as the car, penalised for making the socially optimal choice. The driver is rewarded for making the socially destructive one. The incentive structure is precisely backwards — and invisible, because the question was defined as "how do we move vehicles?" rather than "how do we move people?"

Modena has widened roads six times since 1962. Each time, congestion returned within eighteen months. Six repetitions of the same intervention producing the same failure is not bad luck. It is a system performing exactly as designed — for the wrong objective.

The first discipline is the one most consistently skipped. Not because it is difficult, but because the wrong question feels so obviously right that no one thinks to ask whether it is.

This is the first in a series of six articles exploring the patterns behind Statistics Don't Cough: Six Problems, One City, One Corgi — a book about why intelligent, well-intentioned people get complex problems wrong, and what structured thinking looks like when it's done right. Available now on Amazon.