Short, opinionated pieces on the decisions that move companies — and the quiet, avoidable mistakes that turn good teams into expensive ones.
The most expensive line item on a P&L is the one nobody books — the quiet, compounding cost of conviction that was never actually earned.
Boards reward conviction and punish hedging. The hidden cost shows up two years later, in the decisions nobody remembers framing carefully.
A point estimate is what you produce when the audience is more important than the answer.
A single number is not a forecast. It is a wish, dressed up. Calibrated ranges turn forecasting from a performance into a working tool.
The cheapest insurance product in business is an hour, a whiteboard, and a room full of people willing to imagine they were wrong.
An hour of structured pessimism, scheduled before the commitment, prevents the slow-motion failures that consume the next eighteen months of meetings.
The plan you end up with reads differently from the one your company has been producing. The difference is what we get paid for.
Week by week, what a Bayeseon-run annual planning cycle looks like — from the audit of last year's plan to the board-ready document of bets and odds.
The best forecasters in the world are not subject-matter experts. They are people who have learned, deliberately, how to update.
Amateurs beat the CIA on geopolitical forecasting using three habits. Each one translates directly to the calls a CEO is making this quarter.
The executive most useful to the company over five years is often the one least impressive to the board over fifty minutes.
Boards select for confident communicators because confidence is the only thing they can observe in an hour. The filter produces a systematic distortion.
The calibration crowd has caricatured intuition. The intuition crowd has overgeneralized it. The truth is narrower and more useful than either side admits.
Expert intuition is real and reliable when three conditions hold. Most executive decisions fail at least one of them. The job is knowing which is which.
The red team becomes 'the people who tried to kill the deal' and is, in due course, defanged or disbanded. The fix is structural, not cultural.
Red-teaming sounds good and works badly in most companies because it gets read as adversarial. Four structural fixes separate the working version from the theatrical one.
The journal that no one revisits is a journal that no one writes. The fix is not a better template. It is a ritual.
Decision journals are widely recommended and almost never used at the leadership-team level. The reason is that most templates are built for individuals, not institutions.
An amateur premortem produces a list of weather and regulation. A real one produces a document somebody almost didn't write down.
The premortem is a powerful technique done badly more often than it is done well. Five facilitation failures separate the useful version from the theatrical one.
The retro asks what happened. The post-mortem asks what broke. The decision audit asks the question neither of them does: was the call any good in the first place?
Companies do project retros and incident post-mortems. Almost none audit their decisions. It is the highest-leverage hour a leadership team can spend.
A pricing change is a bet on customer behavior. Treat it like one, and the next change gets easier rather than harder.
Most companies frame pricing changes as forecasts. The framing destroys the ability to learn from the change and makes reversal politically expensive.
The decision to stop is structurally different from the decision to start, and most companies have no process for it.
Every constituency in the company has a reason to keep a dying product alive. The structural drag on the portfolio is enormous — and almost nobody runs the kill decision well.
The credible market-entry case has less numerical density than the dishonest one, and reads more clearly to the board.
The TAM slide is a hallucination, the competitive matrix is a Wikipedia exercise, and the customer interviews were with people too polite to say no. Here is the honest version.
Nobody at an M&A table is paid to honestly stress-test the case. That is the structural problem the second opinion exists to solve.
Every model in a deal room is built to support the deal. Without a structured second opinion, bidders systematically overpay — and the post-merger numbers prove it.
The rep who has missed by 30% for four quarters running is not a 100% commit. The fix is to let the math say so, in writing, before the next forecast lands.
Every quarter, CROs invest in new tooling, CRM hygiene, and MEDDIC discipline. The one fix worth more than all of them is brutally simple and almost never done.
The fastest way to improve a forecast is to stop forecasting from inside the project and start forecasting from the history of projects like it.
One of the most useful ideas in business — and one of the least adopted. The outside view, explained the way a CFO would actually use it on Monday.
The format is short enough to fit on an index card. Most forecasts that cross a CFO's desk fail it on the first line.
There is a specific five-line format a CFO should demand on every forecast. Most companies don't have it. The cultural consequences of installing it are outsized.
The reason your annual plan is wrong is not that the model was wrong. It is that you only run it once a year, against a single number, with no memory of the last five.
Annual plans miss by 20% in either direction with depressing regularity. The fix is structural, not statistical — and it starts with three changes to the operating rhythm.
The minimum-information bet is the cheapest real option in business, and almost nobody buys it.
What is the smallest version of this bet we could make that would teach us most of what we'd need to know to make the big one? Most companies never ask.
Naming the data you don't have is more strategic than mining the data you do.
The data is available for the parts of the decision that are already cheap to be right about. Strategy lives in the un-measurable middle.
Slides reward the visual and hide the assumption. Prose forces the writer to own a verb.
The slide deck is the wrong artifact for strategy. A five-page memo forces the writer to defend the argument and the reader to engage with it.
A strategic plan should read as a portfolio of bets with stated odds — not as a forecast that pretends the future is a known quantity.
Two strategies with the same expected value can have wildly different distributions of outcomes. That is the conversation the board should be having.
Boards have borrowed the word "post-mortem" from engineering and almost none of the practice. The gap is doing real damage.
The engineering post-mortem works because the culture around it is blameless and the artifact is a system fix. The board version is neither.
Confirmation bias and sunk cost get the textbooks. The biases that actually cost companies money do not have catchy names yet.
The dangerous biases in senior decision-making are not the famous ones. They are quieter, more operational, and harder to see from inside the room.
If your quarterly review ends with everyone agreeing the green boxes are green, you held a results meeting and called it a strategy meeting.
Going down the list of initiatives and noting which ones hit is a results report, not a review. The real review asks what you knew, expected, and got.
The single biggest reasoning failure in business is the one that feels most like wisdom: judging the call by how it turned out.
A good decision and a good outcome are different things. The cost of mistaking one for the other compounds across every quarterly review.