The one idea
Judgement is not a personality trait. It is a sequence of questions you can learn to run, quickly, in your head.
The people who seem to always know what to do are not calmer or smarter than you. They are running a checklist so familiar it has stopped feeling like one. This lesson is that checklist, written down.
Work out what the real problem is, what you know, what you don't, and who needs to hear about it — then pick the option you could explain afterwards.
Separate the decision that must be made now from the one that can wait. Protect what is irreversible. Tell whoever is accountable, even when you are also acting.
You are making a decision under uncertainty with an asymmetric cost function. Optimise for the reversibility of the outcome and the quality of the disclosure, not for the probability of being right.
The eight questions
Not sequential steps to complete. A loop you run in about ninety seconds.
1. What is the actual problem?
Almost always different from the problem as presented. The client did not ask "which number is correct" — they asked why two numbers differ. But the actual problem is that they have a board meeting at 2pm and might present a wrong figure. Solve that.
The test: ask what happens if you do nothing. The thing that goes wrong is the real problem.
2. What do I actually know?
Facts, separated from assumptions and from what someone told you. Write them down. Under pressure people reason from "Anand thinks the export ran early" as though it were established, and then act on it.
3. What don't I know, and can I find out in time?
List the gaps, then mark each: findable now, findable later, not findable. This is the question that converts panic into a task list. If the answer exists in a file you can open in ten minutes, that is not uncertainty — it is work you have not done yet.
4. What matters most here?
When several things compete, rank them. A rough order that holds up in most workplaces: safety and legal obligations, then irreversible harm, then money and client trust, then deadlines, then your own convenience.
Explicit ranking is what stops you optimising the deadline while the client presents a wrong number to their board.
5. Who needs to know?
Separate from who needs to decide. Someone can be accountable for an outcome and still be unreachable — they still need to be told, and a message they read in four hours is not wasted.
Ask: who will be surprised later if I do not tell them now? Surprise is the thing to prevent.
6. What could go wrong with what I am about to do?
Specifically: is it reversible? This is the single most useful question in the list.
Reversible things — a draft, an internal message, a caveat, a delay — you can do alone and correct later. Irreversible things — money moving, data deleted, something sent to a client, a public statement, a commitment made — need either authority or a much higher bar of certainty.
When in doubt, choose the reversible version of the action.
7. What is actually my responsibility?
Two failures here, and they are opposite. Overstepping: making a commitment that was not yours to make. Understepping: doing nothing because "nobody told me", when a five-minute action would have prevented real harm.
The line is usually: you can act to prevent harm and to preserve options. You cannot commit the organisation to something.
8. Ask, search, escalate, document, or act?
| Option | When | Cost of getting it wrong |
|---|---|---|
| Search | The answer exists in a document, ticket or record | Time only |
| Ask | Someone nearby knows, and it is faster than searching | Small — a colleague's minute |
| Escalate | Above your authority, or the risk is significant | Low if genuine, corrosive if habitual |
| Document | Always. Alongside everything else. | None. This is free insurance. |
| Just do it | Reversible, inside your remit, and you can explain it | Low by construction |
Most real situations are three of these at once: search for two minutes, act reversibly, escalate in parallel, and document as you go.
The Thursday morning, worked through
The actual problem: not the mismatch. The client may present a wrong number to their board at 2pm. Everything else can wait.
What you know: the two files differ. The board meeting is at 2pm. Your manager lands at 6pm. Anand thinks last week's export predated the corrections — an assumption, not a fact.
What you do not know: which figure is right. Whether the client has already built the deck. Whether anyone promised them the earlier number.
Findable in time: the export timestamp is in the file properties, and the correction log has dates — twenty minutes of work, and it probably resolves which version is right. Not findable: whether the client's deck is already printed. So ask.
What matters most: the client not presenting a wrong number. Above your comfort, above the manager's absence, above looking organised.
Who needs to know: the client, now — they are working to a deadline with bad information. Your manager, by message, even though she is airborne. Anand, because you are relying on his recollection.
Reversible or not: telling the client "don't use either figure until we confirm by 12:30" is reversible and protective. Telling them "the correct figure is 4.2" is irreversible if you are wrong, and it is a commitment you do not have the standing to make.
Your responsibility: you can stop them using a wrong number. You cannot certify which number is right on your own authority.
So, at 9:50 — a holding message:
"Thanks for flagging — you're right that they don't match and I don't want you working from the wrong one before your 2pm.
I'm checking now whether last week's export ran before a set of corrections went in, which would explain the difference. I'll confirm which figure is correct by 12:30 at the latest.
In the meantime, if the deck isn't finalised, I'd hold that slide. Has anything already gone to your board?"
Then message your manager: what happened, what you have done, what you have committed to, and the deadline you set. She reads it at 6pm and finds it handled, with nothing to unpick.
Then do the twenty minutes of work.
Notice you did not solve the data problem first. You spent ten minutes protecting the thing that could not be undone, and you bought a deadline you control.
When you genuinely do not know what to do next
1 of 7Write down what has actually happened, in five lines. The act of writing separates fact from assumption. Most paralysis is caused by holding a tangle in your head that would take four sentences on paper.
Try this
Friday, 4:15pm. You are about to run the monthly payment file — forty-two vendor payments, about eleven lakh rupees, cut-off at 5pm or everything waits until Monday.
Checking it, one vendor's bank account number differs from last month. There is an email from someone at that vendor, two weeks ago, saying their account has changed. The email address looks right. Nobody has verified it.
Your manager is in a client meeting until 5:30. Finance has gone for the day.
What do you do?
Your challenge
Level 3 · IndependentFor two weeks, keep a decision log. Every time you decide something at work without being told what to do — even small things — write four lines: the situation, what you did not know, what you chose, and why.
Then review it. Look for the pattern in what you got wrong. Almost everyone finds one of two: consistently acting when they should have asked, or consistently asking when they should have acted.
You have succeeded when you can name which of those two you are. Knowing your own default failure mode is worth more than any framework, because it tells you which direction to correct when you are unsure.
What people usually get wrong
- Waiting for certainty. It does not arrive. The choice is between acting with 70% information and acting too late with 90%.
- Solving the presented problem. The question asked is rarely the problem to solve. Ask what breaks if you do nothing.
- Treating "my manager is away" as an answer. It explains why you could not ask. It does not explain why you did nothing.
- Doing something irreversible to avoid missing a deadline. Deadlines are almost always the cheaper thing to break.
- Escalating everything. Judgement includes deciding what does not need raising. Escalating routinely trains people to stop reading your messages.
- Acting without telling anyone. Even correct autonomous action damages trust if the accountable person finds out afterwards.
- Not writing down your reasoning. Six weeks later, "it seemed right" is indistinguishable from carelessness.
- Optimising to look decisive. Nobody senior is impressed by speed. They are impressed by "here's what I knew, here's what I chose, here's why."
How someone experienced does it
The thing that most separates experienced people is that they think in asymmetries rather than probabilities. They rarely ask "what is most likely?" They ask "which way of being wrong is survivable?" Late payment versus lost payment. Awkward call versus wrong figure at a board meeting. Once you see decisions this way, most of them resolve in seconds, because the costs are lopsided far more often than the probabilities are.
They also decide at the last responsible moment — not the earliest possible one. Deciding early feels efficient and throws away information that was still arriving. The skill is knowing when the moment is: the point after which delay starts destroying options.
The most senior version is that they make their reasoning visible while deciding, not afterwards. "I'm holding this because a wrong payment can't be recovered and a late one can" lets anyone with more context correct them in seconds. People who announce only conclusions have to be right; people who show their reasoning get help before it matters.
Why good judgement looks like luck from outside
Judgement is invisible when it works. Nobody notices the wrong payment that was never sent, or the client who never presented a bad number. The successes leave no trace; the failures are extremely visible.
So you will not get much credit for it early. The person who prevented a problem looks less impressive than the person who visibly fixed one. Documenting your reasoning is how you make invisible work legible — not to grandstand, but so the decision exists somewhere other than your memory.
It also means you cannot judge your own judgement by outcomes. A good decision can produce a bad result, and a reckless one can get lucky. The only honest self-assessment is: given what I knew, and what I could have found out in the time available, was that reasonable? Judge the process. Outcomes average out over a career; the process is what you can improve.
Prove it
Keep the two-week decision log from the Challenge.
At the end, pick the single decision you are least sure about and write the case for the opposite choice, as convincingly as you can. Not to conclude you were wrong — to find out whether you had actually considered the alternative or just moved fast.
That exercise, repeated a few times a year, is the closest thing there is to deliberately practising judgement.
Keep learning this
Paste this into any AI assistant. It turns the assistant into a tutor that tests you instead of just answering you.
Act as an experienced practitioner who is good at teaching. I have just learned making decisions at work with incomplete information. Assume I am intelligent but relatively new to this — treat me as intermediate level. Work through this in order, and wait for my reply at each step: 1. Ask me 5 questions that test whether I actually understood making decisions at work with incomplete information. Do not reveal the answers yet. 2. After I answer, tell me which parts I got right, which I got wrong, and which I only half-understand. Explain only what I misunderstood — do not re-teach what I already know. 3. Give me one practical challenge based on something I could genuinely encounter at work or in daily life. Do not solve it for me. 4. Evaluate my solution the way an experienced person would judge it, including what a professional would have done differently. 5. Tell me what to learn next, and why that comes next. 6. Give me trustworthy sources for deeper study — prefer official documentation, primary research or standards bodies over blogs and videos. Rules for you: no buzzwords. No motivational filler. Say "I'm not certain" when you are not certain, and tell me which parts of your answer I should verify myself. Clearly separate facts from your recommendations and your opinions.
Become independent at this
Use this when you want a path from where you are to actually good, with checkpoints you can test yourself against.
I want to become independently capable at exercising professional judgement under uncertainty — not permanently dependent on AI, tutorials or step-by-step guides. Design a progression for me with five stages: Beginner, Guided practice, Independent practice, Real-world application, Professional level. For each stage tell me: - what I must know - what I must be able to do without help - the mistakes people make at this stage - one practical challenge - one real project that would prove I reached this stage - one way I can test myself honestly Then tell me the signals that I am ready to move to the next stage, and the signals that I have skipped ahead too early. Keep the theory to the minimum I actually need. Focus on ability I can transfer to situations you and I have not discussed.