• Home
  • About
  • Research Team
  • Projects
  • Talks
  • Publications
  • Blog
  • CV

On this page

  • What Is a Decision?
  • The Hidden Trap: Decisions on Autopilot
  • Examples: Decisions in Supply Chains
  • Examples: Decisions in Public Health
  • Framing the Problem for Making Better Decisions
  • References

What Is a Decision?

decision-making
public health
supply chain
problem-framing
Decisions are something we control — a choice from a known set of options, as opposed to things we merely observe or things that happen to us. This post uses public health and supply chain to illustrate how recognizing decisions — especially the ones made on autopilot — is the first step to making them better.
Author

Bahman Rostami-Tabar

Published

February 26, 2026

It is Tuesday morning. A regional health officer receives a report: vaccine stocks at three districts will run out within 5 days. She has 2,000 doses at the central warehouse and six districts waiting. She could split the doses equally — or prioritise the two districts serving the highest-risk populations. She calls her team, look at the data & models, and weighs the options, and by midday she has signed the distribution order: 800 doses to District A, 700 to District B, and 500 split across the remaining four.

That is a decision.

She had options — different ways to allocate the doses. She was in control — nobody else could sign that order. She committed — the trucks left that afternoon, and redirecting them would cost time, money, and trust. And her choice changed what happened next — some districts got more, others got less, and people’s access to vaccination shifted as a result.

That is all a decision is. The rest of this post is about recognising those moments — and making them better.

What Is a Decision?

A decision is when you choose what to do from different possible options, and your choice changes what happens next.

It is something you control. It could be choosing how much to buy, when to act, what rule to follow, what message to send, or even what information to collect. If you can say, “We will do this instead of that,” and your choice makes a difference, then you are making a decision.

Every decision has three parts:

  1. there are different options,
  2. someone chooses one option,
  3. and that choice affects the future.
TipRon Howard (1966)

“A decision is an irrevocable allocation of resources, irrevocable in the sense that it is impossible or extremely costly to change back to the situation that existed before making the decision.”

NoteA note on terminology

Depending on the field, what we call a “decision” may go by many other names — an action, a choice, a control, a design, an option, or a task — but they all refer to the same thing: something we are in control of.

The Hidden Trap: Decisions on Autopilot

Here is the part that catches everyone — experts included.

We often make decisions without realising we are making them. We follow last year’s protocol, the usual supplier, the default treatment guideline, the same route — on autopilot, without ever stopping to ask: is this still the right choice?

But autopilot is still a choice. Not deciding is a decision. Doing what you always did is a decision.

WarningThe key insight

You cannot make a better decision until you first recognise that a decision is being made.

That single insight — just noticing that you are at a fork in the road — is where improvement begins.

Examples: Decisions in Supply Chains

Supply chains involve hundreds of interconnected decisions, many of which are automated or habitual — making them easy to miss. What makes them decisions is that someone chose them, and someone can change them.

Area Decision What Is Being Controlled?
Inventory replenishment Choose the exact date to place a replenishment order and how many units to request, given current stock, expected demand, and supplier lead time When the order is placed and how large it is
Safety stock Choose the minimum stock level to maintain at each warehouse or clinic before triggering a new order (the “safety stock”) — this directly sets the buffer against supply disruptions Stock levels held at each location
Distribution Choose how to split a limited shipment of vaccines or medicines across competing regions — which regions get more, which get less, and in what sequence How scarce doses are allocated across locations
Routing Choose which roads or transport corridors each delivery truck follows, and which warehouse is responsible for fulfilling each clinic’s order Physical paths taken and assignment of supply sources to destinations
Supplier selection Choose which pharmaceutical companies or distributors to buy from, based on price, reliability, cold-chain capability, and lead time — a decision that is costly to reverse once contracts are signed Who supplies the goods and under what terms
Facility planning Choose the specific cities or regions where cold-storage warehouses will be built or rented — this determines which areas can be served quickly and which face longer delays Physical location of storage infrastructure
Network design Choose the total number of warehouses to operate, where each is located, and which regions each one serves — this shapes every other supply chain decision that follows The overall structure of the supply network

Examples: Decisions in Public Health

Public health is full of decisions at every level — from choosing a clinical protocol for a single patient to setting policy for an entire population. What makes them decisions is that someone is in control of them: they can be changed, debated, and improved.

Area Decision What Is Being Controlled?
Disease prevention Choose whether to launch a vaccination campaign at all, and if yes, set the start date, target population, and delivery method (clinic, school, mobile unit) Whether, when, how, and for whom to act
Resource allocation Choose how many vaccine doses to send to each region, in what order, and which regions get priority when supply is limited Distribution of scarce vaccines across competing needs
Screening policy Choose which population groups to test (e.g. high-risk vs. all adults), how frequently (annually vs. every 3 years), and which test to use (cheap/fast vs. accurate/expensive) Who gets tested, how often, and with what tool
Public communication Choose to broadcast a disease risk warning on TV rather than social media, because the at-risk population is older and less active online Which channel carries the message and reaches the right audience
Information campaigns Choose the specific message to send (e.g. “wash hands for 20 seconds” vs. “avoid crowded spaces”) and define exactly who receives it (all adults vs. parents of young children) The content of the message and who it is aimed at
Intervention timing Choose the specific date to activate a health alert, enforce a mask mandate, or close schools — too early wastes trust, too late costs lives When the action takes effect
Data collection Choose which surveillance systems, registries, or research studies to fund and build — this shapes what health information will be available for all future decisions What gets measured and therefore what can be known

Framing the Problem for Making Better Decisions

To make a better decision, you must first frame the problem clearly. That means answering three questions in the right order.

NoteStep 1 — Start with the performance metrics: What does “good” mean?
  • What are we trying to improve? (fewer infections, lower cost, faster delivery)
  • How will we measure success?

You start here because you cannot choose well if you do not know what “better” means.

NoteStep 2 — Identify the decisions: What can we control, and who controls it?
  • What choices can be made?
  • Who is allowed to make those choices?

These are the “buttons” we can press — like how many vaccines to send, when to act, or which supplier to use.

NoteStep 3 — Identify the sources of uncertainties: What don’t we know?
  • What might happen that we cannot control? (e.g. disease spread, demand, delivery delays)
  • Does this uncertainty change which decision is best?
  • Does this uncertainty strongly affect performance?

If uncertainty matters, we can collect better data, add safety buffers, or make flexible plans.

TipIn short
  1. Decide what success means.
  2. List what you can control.
  3. List what you cannot control.
  4. Make choices that work well even when surprises happen (e.g. use Sequential Decision Analytics)

References

  • Bridging Decision Problems

  • Framing the Problem (Bridging Decision Problems Book 2)

 

Made with ❤️ and Quarto