Judge the Call, Not the Dice

One framework for choices at work: hitting a sales number, launching a product, getting a team to use a new tool. It starts with what you value and ends with what happened, and it keeps luck in its place.

Part 1

The bad week

Picture a quarter where your team did everything right. You talked to customers, built a solid pipeline, and ran the plan. Then the sales number came in below target.

The tempting conclusion is that the plan failed. Most teams drop the plan right there. A few weeks later they pick a more aggressive one, and that one fails too.

Now picture the opposite quarter. You skipped the customer work and pushed a rushed promotion, and the number came in above target. The tempting conclusion is that you got away with it. So you do it again.

Both conclusions make the same mistake. They judge the choice by a single result. Poker players have a name for this: resulting. The book Thinking in Bets made the word popular outside poker.

Resulting punishes good calls that got unlucky, and rewards bad calls that got lucky.

Over time, resulting teaches you the wrong lessons. You drop sound plans after one bad quarter and keep reckless ones after one good quarter. This essay is a way out of that trap. It works the same way for one person and for a whole team.

Part 2

An outcome has three ingredients

Start with a simple sum. What happened is the result of three things mixed together.

The decision is the choice you made with what you knew at the time. Execution is how well you carried it out. Luck is everything else: the competitor that launched the same week, the market that turned, the key engineer who left mid-project.

Only the first two are yours. That gives four possible stories for any result, and only two of them are fair to the decision.

Two words keep the stories straight. The goal is what you intend, and you set it before you start. The outcome is what happens, and you only see it after. Success means the outcome matches the goal.

That sounds obvious. It stops being obvious when the goal was never written down. Then the goal quietly becomes whatever happened, and every outcome counts as success.

The same test scales up from a single goal to a whole life. Here is my own definition.

A successful person is someone whose life keeps landing close to goals they wrote down, without trading away what they value to get there.

Both halves matter. The first half asks whether you hit what you aimed at. The second half asks whether you stayed yourself on the way.

Notice what the definition leaves out. Someone who wins the lottery hit a jackpot, but they never wrote it down as a goal and nothing they chose made it happen. Nobody calls them successful. We call them lucky, and that everyday instinct is the whole point of this essay.

Now picture the opposite case. In 1994, a young finance executive noticed that use of the web was growing about 2,300 percent a year. His hypothesis was simple: the internet was going to be big, and he wanted to build something there. He left a good job, made a plan to sell books online, and told early investors there was roughly a 70 percent chance they would lose their money.

That founder was Jeff Bezos, and the company was Amazon. He did not rely on luck. He relied on a plan, and he had the grit to work it through years of ups and downs. When the facts changed, he updated the strategy and the decisions, not the goal. He knew the risks going in, and he chose a bet whose misses he could survive.

He could have failed. The plan would still have been a good one, made by a person who knew how to make good bets. Over a lifetime, a person like that keeps placing good bets, and good bets keep landing close to the goal. That is why one failure does not undo a successful person, and one jackpot does not make one.

The rest of this essay is about earning both halves of that definition.

Part 3

What a good decision is

If you cannot judge a decision by its outcome, what do you judge it by? By the whole range of outcomes it could produce.

Every choice opens up a spread of possible futures. Some land on the goal. Some land near it. Some miss badly. A good decision gives you the best expected shot at landing close to your goal, and it keeps the misses survivable.

Economists call this maximizing expected utility. The formal idea goes back to the 1944 book Theory of Games and Economic Behavior. The everyday version is simpler. Do not only ask how likely you are to hit. Ask what the misses look like.

A deep discount at the end of a quarter shows why the misses matter. It can hit the number fast. But its misses are ugly: deals pulled forward from next quarter, buyers trained to wait for the discount, and margin that does not come back. Building a steady pipeline has a smaller chance of a dramatic month. Its misses are mild, and you can recover from them.

A good decision is the best shot at landing close, with misses you can survive.

The same test works at work. Rolling a new tool out to every team at once might win fast. If it fails, everyone remembers the failure and the next tool starts behind. Starting with one team that wants it has a smaller upside in month one. Its worst miss is a quiet lesson, not a public one.

Part 4

Goals in layers, not cop-outs

A single number makes a brittle goal. Hit it and you may have won the wrong way. Miss it by a little and it feels like total failure. A better goal has three layers.

Layers can look like an excuse. "I only need to hit the floor" sounds like lowering the bar. So each layered goal has to pass three tests.

1. Can it still fail? If no outcome counts as a miss, it is not a goal.
2. Was it set before? A floor chosen after the results are in is just moving the goalposts.
3. Are you still chasing the upside? The stretch only counts if you work for it.

A cop-out goal fails the first test. "Explore AI opportunities." "Improve collaboration." These are fine intentions, but you cannot tell whether you are missing them. A goal you cannot miss cannot guide you.

There is one more gap to watch. You can hit a goal exactly and still miss what you actually cared about. A sales team can hit its number by discounting away the margin. A team can hit an adoption number by forcing logins that nobody uses for real. This is the values and goal gap, and the guardrail is there to close it.

Part 5

The whole chain

Put the pieces in order and you get a chain. It runs from what you care about to what happened.

Read the chain from left to right.

1. Values and identity. What you care about, and who you are trying to become. For a team, this is its culture.
2. Goal. The floor, the stretch and the guardrail, set before you start.
3. Decision. The choice, judged by its spread of outcomes. A team writes it in a decision journal, so later reviews judge the call and not the memory of it.
4. System design. How the decision becomes automatic: cues, surroundings, defaults. For a team, this is process design: workflows, tools and incentives.
5. Habits. Running the system day after day. For a team, these are its routines and operating rhythm.
6. Progress. Early signs that you are getting closer.
7. Outcome. What actually happened.

Luck does not act on the first four steps. You choose your values, your goal, your decision and your system. Luck acts from the habits onward. It interrupts routines, bends progress and lands on the outcome.

The chain also loops back on itself. Each time you carry out a habit, you cast a small vote for the kind of person you are becoming. Atomic Habits makes this point well. The manager who writes down every big decision starts to see themselves as someone who decides with care. That identity then makes the next written decision easier.

The same book argues that systems matter more than goals. The chain keeps the goal anyway. Systems drive progress, but the goal defines what counts as progress. Without it, a busy system can run in the wrong direction for a long time.

Part 6

Where habits fit

A decision with no habit behind it is a wish. Atomic Habits gives four laws for building a habit: make it obvious, attractive, easy and satisfying. They translate cleanly from one person to a whole team.

Teams add problems a single person does not have. Every routine needs a clear owner, or it belongs to nobody. Incentives can pull against the habit you want. Old routines compete for the same hours. And decisions vanish unless someone writes them down.

Getting a team to adopt a new tool is habit formation at scale.

That reframe changes the plan. A launch email is a cue, not a habit. Adoption grows when the tool sits inside the workflow, a respected person uses it in the open, the first use takes minutes, and the first win gets noticed.

Part 7

Three kinds of measures

The chain needs measures, and it needs three different kinds. Each one answers a different question.

A good progress measure

A good progress measure predicts the outcome, and it moves faster than the outcome does. You can influence it. It is cheap to measure and hard to game. It has a starting point and a clear direction. Two or three per goal is plenty, and each one needs a guardrail beside it.

The guardrail matters because any measure that becomes a target starts to bend. Economists know this as Goodhart's law. Weekly active users can be pushed up by forcing logins while real use stays flat. The guardrail, "no forced logins," catches that.

A good input measure

A good input measure mirrors the plan directly, and it is fully in your control. It tracks consistency more than volume. It captures quality, not just whether you showed up. And it is nearly effortless to log, or nobody will log it.

Three traps catch most people.

1. Treating inputs as progress. Ten customer calls a week is the work. It is not proof you understand the customer better.
2. Streak perfectionism. A broken streak makes people quit. Aim for about 80 percent of days each week, and never miss twice in a row.
3. Counting what is easy. Calls are easy to count. Whether you learned anything from them is what matters.

Part 8

Two loops, one diagnosis

Measures only help if you look at them on the right rhythm. The chain has two review loops, and they run at different speeds.

Keeping the loops apart prevents two classic mistakes. The first is changing the plan when the real problem is the habit. The second is fixing the habit over and over when the plan itself is wrong.

The figure turns into a short diagnosis. High inputs with flat progress point to a decision problem: you are doing the work, and the work is the wrong work. Low inputs point to an execution problem: the plan was never really tried. If neither fits, look at luck and noise before you change anything.

Part 9

Three worked examples

A framework earns its keep on real choices. Here are three from work.

Hitting a sales number

The value is a business that grows for the right reasons. The goal is a specific number by the end of the quarter, with healthy margins. Revenue is only a stand-in for a healthy business, which is why the margin belongs in the goal itself.

The decision is building pipeline over a last-minute discount, because its misses are smaller. The judging happens over several quarters. A single quarter is noise. One big deal slipping a week can move it in either direction.

Launching a new product

The value is building something customers actually need. A breakout hit would be a wonderful upside. It is not the goal.

The goal is to learn fast what customers will pay for, and to keep the option to scale if the signal is real. The dangerous miss is a big launch that burns the budget, tires out the team, and teaches the company to distrust new ideas.

So the decisions follow. Customer pull sets the pace, not the launch calendar. Run several small bets early. Judge progress on what you learned and how often customers come back, not on launch-day buzz. And ask one question often, with real honesty: is this goal the customer's, or ours?

Getting a team to use a new tool

The inputs are weekly office hours, starter templates that take two minutes to use, and a monthly check-in with each team lead. The progress measures are weekly active users, repeat use, work actually finished inside the tool, teams asking for new features, and people starting without being reminded.

The coaching matters as much as the plan. Let teams struggle a little before stepping in. Praise the new workflow, not the usage number. End each training session on a win, so the next one starts easier. The right mix also depends on the team. A team that is new to the work and a team of veterans need different amounts of support.

The same chain runs under all three examples. Write the goal before you start. Choose by the spread of outcomes, not the best case. Build a system so the habit runs itself. Measure inputs weekly and progress monthly. When results disappoint, ask whether it was the call, the work or the dice.

Judge the call, not the dice. Then fix the part that actually broke.

Sources

Where the ideas come from

• Thinking in Bets (Portfolio, 2018), on resulting, and on judging decisions apart from their outcomes
• Atomic Habits (Avery, 2018), on systems over goals, habits as votes for an identity, the four laws of behavior change, and never missing twice
• Theory of Games and Economic Behavior (Princeton University Press, 1944), for the formal idea of expected utility
• The 4 Disciplines of Execution (Free Press, 2012), on lead measures that predict and can be influenced, versus lag measures that report the result
• Academy of Achievement, interview with Amazon's founder (2001), on the 2,300 percent growth figure and the decision to leave a stable job; the 70 percent odds he gave early investors come from his later public talks
• Goodhart's law, from economic policy writing of the 1970s, on measures that stop working once they become targets