Leadership is all about making decisions (or just as importantly not making decisions). Decision-making is hard because there are many types of decisions, and just as many tools for making decisions. It’s easy to default to a one-size fits all model of decision-making that’s slow and laborious, or flip the other way and become a modern-day Dice Man.
In this post, I’ll try to help define the variables of a decision, show some tools you can use for different types of decision, and close out with some general ideas about how to change an expensive decision to a cheap one. Hopefully this’ll mean fewer committees deciding about the colour of the bike shed.
The anatomy of decision difficulty
Decisions exist on a spectrum of difficulty, and that difficulty comes from specific, identifiable sources.
Easy decisions share common characteristics:
Clear criteria: You know exactly what you're optimizing for
Obvious winners: One option is demonstrably better than alternatives
Low stakes: The consequences are manageable regardless of choice
Adequate information: You have enough data to choose confidently
Single stakeholder: You're the only person affected (or the primary decision-maker)
Think about choosing a coffee shop when you’re in an unfamiliar town. You know your preferences (criteria), can quickly rule out options you dislike (nope, we’re not going to Costa), won't suffer dramatically from a mediocre coffee (just go somewhere else!), can check reviews online (adequate information), and it's your choice (single stakeholder).
Hard decisions, by contrast, involve:
Competing values: Multiple important criteria that conflict with each other
Uncertain outcomes: The consequences are genuinely unpredictable
High stakes: Getting it wrong has significant costs
Information gaps: Key data is missing, contradictory, or unreliable
Multiple stakeholders: Many people are affected with different preferences
Irreversibility: The choice can't easily be undone
Consider deciding whether to acquire a competitor. This involves competing values (growth vs. financial stability), uncertain outcomes (will the integration work?), high stakes (could make or break the company), information gaps (internal competitor data), multiple stakeholders (shareholders, employees, customers), and significant irreversibility (acquisitions are expensive to undo).
Daniel Kahneman's work shows that our brains use different systems for different types of choices. System 1 handles easy decisions automatically and intuitively. System 2 kicks in for hard decisions, requiring deliberate, analytical thinking. The problem? We often use System 1 for decisions that deserve System 2 attention, and System 2 solutions for choices that should be quick and intuitive. The trick is matching System 2 effort to decisions that warrant it, and protecting simple, reversible calls from unnecessary ceremony.
When you’re faced with a decision, your first challenge is to work out what kind of decision you’ve got:
Impact - is it a mundane day-to-day task? Or a strategic change of direction?
Reversibility - is it a Type 1 or a Type 2 decision? Type 1 decisions are almost impossible to reverse (a one-way door). Type 2 decisions are easy to reverse.
Information - are you making this decision based on concrete data, or is the data itself uncertain? Have you made decisions of this type before, or is this a novel decision?
Stakeholders - is this decision just for you or does it affect a wider group of people? This is subtly different to impact; consider how much alignment the affected people have with the decision.
Feedback latency - once you’ve made a decision, how soon will you know whether it was good?
Decision Archetypes
Once you know the “shape” of your decision, you can start to find a tool to help you make the decision. Tools are important because they systematically engage System 2 thinking and help you tackle cognitive bias.
Routine decisions are day-to-day operational decisions. The main challenge here is not to overcomplicate them, but also to acknowledge that sometimes System 2 needs to be engaged!
Automate the boring bits - For example, use a linter to format your code or use a service that brings lunch in.
Checklists - walk through the key bits of process in any complex procedure
Crisis decisions are those you have to make quickly, with limited information, high impact and multiple stakeholders. This might be something like responding to a security breach (or dealing with a pandemic). There’s no time for extensive analysis. Pre-decide roles and thresholds (runbooks!) to reduce time-to-action.
OODA - Observe, Orient, Decide, Act. Repeat! Information is changing at pace in a crisis.
A strategic bet is a high impact, often irreversible decision with unclear outcomes and multiple stakeholders. There’s almost an endless supply of messy data, but you have to balance extensive analysis against decision speed.
WRAP - Widen your options, Reality-test your assumptions, Attain distance, Prepare to be wrong. These are decisions where investing in making the most right choice is worth the effort.
An optimization decision is a choice between some courses of action. Do I invest in A, B or C? It’s often a choice of multiple “good” options, and the particular challenge is avoiding over-analysis of improvable decisions. The challenge here is agreeing the few criteria that matter most! Prefer coarse weights and ranks and avoid spurious decimals!
Decision matrices - identify the criteria you care about and rate the performance of each one.
A/B testing - run an experiment to determine whether A or B is better - use data to guide the decision.
Exploration decisions are open-ended of the type “what do we do next” in an uncertain environment. These are often learning-focused, experimental and reversible decisions made with very uncertain data.
Design Thinking - Deeply understand the problem first. Co-evolve solutions with stakeholders.
Lean startup - state the hypothesis, run the experiment, pivot or persevere.
Innovation decisions are the long-term bets on ideas that don’t have precedent. They’re high-uncertainty, often high-impact, and rarely reversible in the traditional sense but you can structure them to preserve flexibility. Think of them as “real options”: you place small bets now (time, money, attention) to keep open the possibility of bigger commitments later. Innovation decisions are less about optimizing the present and more about shaping the future. Treat each stage as a two-way door, only the final scale-up becomes one way.
Real Options – invest in small, staged experiments that give you the right (but not the obligation) to scale later.
Portfolio Thinking – treat innovation like venture capital: spread your bets across multiple ideas, knowing some will fail, a few will pay off, and you can double-down on the winners.
Value trade-offs are those with conflicting priorities and many stakeholders. These might be the hardest of all, such as making redundancies or not. These are not just about data or reversibility, they’re about humans and how they are affected.
Stakeholder mapping - A stakeholder map helps you understand those affected by a decision, and will help you understand who to involve in the decision-making process.
Ethical frameworks - What values does your company live by? If they’re real, then they should be used in decision-making (but see oil companies say “environmental” and destroy the environment, finance companies say “we’re ethical” and invest in war and so on, or “do no evil” and then …).
Making Decisions Easier
As well as choosing a decision, you can often choose to simplify the criteria to make decisions easier.
Clarify the problem - The better you understand the problem the easier the decision.
Reduce the scope - Don’t try and boil the ocean! Instead of committing to a strategic bet, reframe as an exploration decision.
Allow Reversibility - Can you break a large decision into a series of two-way tests?
Improve feedback loops - We often make bad decisions and never know because the feedback loop never gets closed (see Why speaking to users is a good idea).
Speak to stakeholders - Sometimes we panic about decisions where the stakeholders are already aligned. You might use nemawashi style techniques to gauge feedback from stakeholders prior to taking the decision.
Delegate - Is this decision complex because you’re not the right person to make a decision?
The key here is that if a decision feels heavy, use these levers to make it a lighter decision.
Meta-Decision Making
How do we decide how to decide? Do we decide by consensus, by majority vote or by single-point accountability? Decision-making isn’t just about making good decisions. Sometimes the mere act of making a decision has second order impacts.
In a model where the leader decides everything, it’s speedy but it disenfranchises the team. At the other end of a spectrum, if everyone decides everything then it’s inclusive (a good thing!), but it’s slow and exhausting.
There’s a few ways to tackle this process.
Jurgen Appelo introduces the 7 Levels of Delegation which makes explicit about how much autonomy a decision carries, from “Tell” through to “Delegate”.
In the Advice Process model, anyone can make a decision but they must seek advice from those affected and those with expertise.
Amazon have an explicit “Disagree and Commit” model. Prior to a decision, we can hear the reasoning from all sides, but after a decision is made folks align behind it even if they weren’t directly involved.
In all models, decision logs are helpful because they share the reasoning and outcomes, so people understand why even if they weren’t directly involved (see Architecture Decision Records and the theme of making expert thinking visible in Supporting your mid-level engineers).
My heuristic is simple.
If the team will live with the consequences more than you will, they should have a meaningful role in the decision
Putting it all together
So here’s my meta-process for making decisions.
Diagnose – Understand the type of decision you have by looking at the impact, reversibility, information available, stakeholders involved and feedback latency.
Match – Pick the archetype that best fits that shape (crisis, strategic bet, optimization, exploration, innovation, value trade-off, delegation) and use the tool that matches. Don’t bring a spreadsheet to a firefight, or OODA-loop your way through a strategic bet.
Simplify – Make the decision lighter if you can. Can you:
Reduce the scope?
Break it into reversible steps?
Tighten the feedback loop?
Clarify the problem statement?
People – Decide how you’ll decide. Who needs to be involved? What level of delegation is appropriate?
Heuristic: if the team will live with the consequences more than you will, they should have a meaningful role in the decision.
Act & Review – Make the call, document your reasoning (an ADR-style note is enough: date, owner, type, assumptions, expected signals, review date), and revisit when feedback comes in. The act of writing sharpens the decision, builds collective memory and shares expert thinking.
The point isn’t to find perfect answers. Perfect doesn’t exist. The point is to make better bets, more systematically, and build a system where future decisions get easier because your team has a shared language, process, and archive of lessons learned.
I was pointed at https://thedecider.app/ which looks like a fantastic tool to automate this blog post and save reading it :)