The reasoning underneath every good call: how to think clearly, weigh evidence, handle uncertainty and read the economics before you decide.
Three engines drive almost every decision you make: building an answer up from scratch, reaching for a fast rule of thumb, or borrowing the shape of a problem you've solved before. Knowing which one you're running, and which one the moment actually needs, is most of the skill.
Three ways of drawing a conclusion, proving it, generalising to it, or explaining your way to it. Knowing which one you are doing is the difference between a confident decision and a confident mistake.
Most bad decisions in a meeting aren't lies, they're arguments whose shape is wrong, dressed in confident language. Here's how to see the shape.
A mental model is a compressed theory of how some part of the world works. The leaders who decide best don't own one good model, they keep a latticework of them, borrowed from many disciplines, and reach for whichever one the problem actually needs.
Three small pieces of school maths quietly decide most of the calls you make: what overlaps, what's likely, and how many ways it can go. Get them straight and a lot of confident nonsense stops sounding convincing.
Most problems that won't stay fixed aren't bad-luck events, they're produced by a structure that keeps generating them. Systems thinking gives you the vocabulary to see that structure, and the leverage to change it.
Good outcomes flatter bad decisions and bad outcomes punish good ones. A decision journal, plus a little calibration practice, lets you tell the two apart, so your team actually learns from experience instead of just collecting it.
Most "we're too slow" problems aren't speed problems, they're ownership problems. Two simple tools, RAPID and the one-way/two-way door, fix the two questions teams keep getting tangled in: who decides, and how much process the decision deserves.
The single most useful question to ask before any decision isn't "what's the right call?", it's "can I walk this back?" Get that one right and you'll decide faster on the small stuff and more carefully on the stuff that counts.
When you can't know the answer yet, the smart move often isn't to commit or to freeze, it's to buy the right to decide later, cheaply, once the fog clears a little.
Four ways a group can decide together, and a simple rule for choosing between them, grounded in two centuries of maths about when crowds get it right and when they get it badly wrong.
Three tools that look the same and aren't: one weighs whether to act, one weighs what to give up inside an option, and one weighs the door you quietly close. Used together, they stop confident motion in the wrong direction.
The moment you delegate, you hand a goal to someone whose interests don't perfectly match yours, and the gap between what you want and what they're actually paid to do is where most management surprises live.
Some of the most expensive things your team produces never show up on an invoice. Learn to see the spillovers, and to decide whether to price them, fence them, or govern them.
Every time you choose to hire instead of contract, build instead of license, or own instead of partner, you are answering one of economics' oldest questions, why do firms exist at all? Here's the theory that answers it, and how to use it on your next make-or-buy call.
Some products get better the more people use them, and a few of those connect two different crowds who each need the other to show up first. Understanding which kind you have changes almost every decision you'll make.
Two sets of decisions you don't control, what the central bank does to interest rates, and what the government does to tax and spending, reach almost every line of your plan. Here is how the signal travels, and what to do before it lands.
Most of your hardest calls, who to hire, who to trust with the budget, which supplier is bluffing, are really one problem: the other side knows something you don't. Information economics names that problem and, better, tells you what to do about it.
Most leaders don't lack information about their market, they lack the questions that turn it into a decision. This is how to gather competitive intelligence that actually moves you.
Two people can both say "the evidence shows" and mean wildly different things, one a controlled trial across ten thousand people, the other a story they heard at a conference. Knowing which is which is a leadership skill, not an academic one.
You don't need an ethics board to learn about your customers and your team. But the moment you collect data on real people, you've inherited the same three obligations that govern medical trials, and breaking them quietly is how good companies end up in the headlines.
Three skills that turn a pile of half-agreeing, half-contradicting inputs into one decision you can actually defend, and the discipline to know which clashes to chase.
People are unreliable narrators of their own lives. Good interviewing and a little fieldwork close the gap between what they tell you and what they actually do, which is usually where the decision lives.
People don't buy a quarter-inch drill because they want a drill, they want a quarter-inch hole. Jobs-to-be-Done is the discipline of finding the hole before you fall in love with the drill.
A metric is not a thermometer you read, it is a lever your team grabs. Design it badly and people will optimise exactly what you asked for, and nothing you meant.
Picking a chart isn't a styling decision, it's an argument about what the numbers mean. The research on how people read graphs tells you which chart wins, and which one quietly misleads your team.
Any number that moves over time hides three things at once, a direction, a rhythm, and noise. Pull them apart, and a foggy chart becomes a decision you can actually defend.
A single average retention number hides the only thing worth knowing. Split your users into cohorts, watch each curve over time, and the question stops being "how many came back?" and becomes "for whom is this genuinely sticky?"
Regression is the maths behind nearly every "if we change X, what happens to Y?" question you'll ever ask a dashboard. Here's how to read one without being fooled by it.
Before a number can lie to you, it has to be the wrong type. Knowing which of the four kinds of data you're holding is the cheapest mistake-prevention a leader can buy.
Most leaders are handed two bad options when a new technology arrives: panic-adopt, or ignore it until a competitor forces their hand. Horizon scanning is the discipline that gives you a third, watch early, judge soberly, and commit when the evidence is there, not when the hype is loudest.
Governing AI isn't about writing a policy nobody reads. It's about deciding, before anything goes live, who is accountable, what the system is allowed to do, and how you'll know when it's wrong.
Four words that decide whether the number on your dashboard is something you can bet a budget on, or a confident guess wearing a chart. Here is what each one means, and the moves that make the difference.
Warehouse or lakehouse, star schemas, a semantic layer, data mesh, master data, the words your data team uses every day, translated into the few ideas that actually change the decisions you sign off on.
A machine-learning model is the only thing your team ships that can be correct on launch day and quietly wrong six months later, without anyone touching the code. MLOps is the discipline of noticing before your customers do.
Automation replaces a task. Augmentation makes a person better at one. Knowing which you are buying decides where AI creates value, and where it quietly destroys it.
Leaders Loop
