In this article
Before you start
This page is one question inside the behavioral interview questions guide. The broader skill of shaping any answer is in how to answer interview questions.
Why interviewers ask this
A weak answer to this question sounds like either recklessness or paralysis, and the interviewer cannot tell which. Behavioral interviewing is based on the idea that past behavior predicts future performance, and openers such as "Tell me about a time when…" ask for a real past event. Candidates report versions of this question about a quick decision with limited information; no public source shows which companies ask it.
The question separates people who guess from people who know what they know. Amazon's published Bias for Action principle says "Speed matters in business" and adds "We value calculated risk taking", so even a company known for speed asks for calculation. A weak answer says "I went with my gut." A strong answer says what was known, what was missing, why waiting cost more, and how the error would show. The Amazon leadership principles page covers the wider set.
Decision with incomplete information from first principles
- Decision with incomplete information
- A decision with incomplete information is a choice made before some facts that could change it are known, because getting those facts would cost more time, money or opportunity than the choice is worth waiting for, so the decision maker accepts a bounded risk.
- Tripwire
- A tripwire is a check agreed before you act, such as a metric threshold or a date, that tells you early whether a decision is going wrong, so that you can reverse or adjust it before the cost grows.
In plain words
Think of choosing a route when the traffic map has not loaded. You know the roads. You do not know today's jams. You pick a route, and you watch the clock. If you are late at the first junction, you switch. The analogy breaks in one place. A driver can always turn back, but some work decisions cannot be undone.
Here is the same idea in one precise sentence. A good answer shows a decision made on partial facts, sized to how reversible it was, with a way to notice and fix an error.
The interviewer is listening for four things. They want the gap, the door, the check and the result. The four steps below take them one at a time.
Step 1: Name the gap in what you knew
The answer names what was missing before it names the choice. That order matters because the gap is what makes the story about incomplete information at all.
Say what you knew, what you did not, and why you could not wait. This is your Situation and Task, and it should be short. Give the gap in plain facts or in a fraction. In the illustrative scenario used on this page, a person has three of five key signals, which is 60% of the information, ten points under the rule of thumb of about 70% in Bezos's 2016 letter.1
Input: the situation. Output: one sentence that states the unknown and the deadline. "I did not have the logs, and customers were failing at checkout" is a gap. "It was a busy time" is not.
Step 2: Judge the door: can the decision be undone?
The reversibility of the decision sets how much evidence is enough. A decision you can undo is called a two-way door, and one you cannot undo is a one-way door.
Bezos's 2015 letter says "Type 2 decisions can and should be made quickly by high judgment individuals or small groups." His main warning is that large organizations use the heavy process on reversible choices, which causes slowness and too little experimentation. A footnote in the same letter names the reverse error, using the light process on a one-way door.
Amazon Web Services (AWS) guidance says to act on a two-way door informed by initial data, so speed is not acting blind. In the interview, say honestly which kind of door it was, and why.
Step 3: Set a check that tells you if you were wrong
A check tells you early if the decision was wrong, and it is the part answers most often leave out. Bezos wrote that you need to be good at "quickly recognizing and correcting bad decisions" whichever way you decide.
Name the check: a number to watch, a date to review, a person to ask. It can be a tripwire, such as "if errors did not fall within thirty minutes, I would reopen the question." Then say what you would do if it fired. This answers the follow-up trap before the interviewer sets it.
The 70% heuristic from Step 1 does not replace the check; it only tells you when to stop waiting.
Step 4: Commit, act, and report a quantified result
The result is quantified and the lesson is stated. Once the choice was made, say what you personally did. Amazon's principles say that once a decision is determined, leaders "commit wholly".
MIT's STAR guidance says to highlight quantifiable results and to describe the actions that you personally took. State what the check showed, what changed in a number, and what you would repeat or do differently. A result need not be a triumph. A clear lesson from a partly wrong decision shows the same skill.
You now have the four moves. The next part shows how long each one should take when you say it aloud.
Step-by-step description
- Say what was missing before you name the choice.
- Reversibility sets how much evidence is enough.
- A check tells you early if you were wrong.
- Give a quantified result and the lesson.
- The answer names what was missing before it names the choice.
- The reversibility of the decision sets how much evidence is enough.
- A check tells you early if the decision was wrong.
- The result is quantified and the lesson is stated.
How long each part of a 90-second answer should take
Action is the longest part of a STAR answer, so most of the time goes to how you decided. Northwestern suggests Situation 15%, Task 10%, Action 50% and Result 25%. MIT suggests Situation 20%, Task 10%, Action 60% and Result 10%. The two differ, so treat the split as a range.
In an illustrative 90-second answer, Action takes 45 to 54 seconds, depending on the split. The result takes 9 to 22.5 seconds.1 Steps 2 and 3 live inside Action, so they deserve most of that time. MIT also suggests preparing three to five stories, so one can be chosen for this question.
Text description of the diagram
- In a 90-second answer, Action takes 45 to 54 seconds, depending on the split.
- The result takes 9 to 22.5 seconds.
Knowing it is half the job. Saying it is the other half.
Interview Clozo is an AI interview copilot for real interview rounds: it hears the question and puts a strong answer on your screen, and you say it in your own words. See the features or how it works.
Start Free Trial →The Staff-level answer
Tell me about a time you had to make a decision without all the information.
Our checkout error rate jumped on a Friday and I had to choose between rolling back a release and waiting for logs. I had three of the five signals I wanted, and the rollback was reversible within a day, so I treated it as a two-way door and rolled back. I also set a check: if errors did not fall within thirty minutes, I would reopen the question.
They fell by half, so the rollback stopped the regression. I ordered the missing logs afterwards; they showed the remaining errors came from a config change, which I fixed the next week. Next time I would act just as early, but I would write down what I did not know before I acted.
The figures in this answer (three of five signals, a one-day rollback, thirty minutes) are illustrative.1 The spoken structure is the gap, the door, the check, the result, then the lesson. What makes it Staff-level is that the answer sizes the decision to its reversibility and shows how the error would have surfaced.
A weak version of the same story (illustrative)
"Checkout was broken one Friday, so I went with my gut and rolled back. It worked out." This answer never says what was missing, whether the rollback could be undone, or how the error would have shown. It gives no number and no lesson, so the interviewer cannot tell skill from luck.
Example in a product role (illustrative composite)
A product manager had survey results from three of five customer segments before a launch date that could not move. She confirmed the feature could be switched off per customer, launched to the three segments, and set a rule: if support tickets from them rose sharply in a week, pause. Tickets stayed flat, so she extended to the other two segments and reported adoption against the goal.
Example in an operations role (illustrative composite)
A shift lead had to pick a carrier while one tracking feed was down. A wrong pick would cost one late delivery, so he chose the carrier with the better record, asked for a status call after two hours, and switched the remaining orders when the call showed delays. He reported the late orders avoided.
Example when the door was one-way (illustrative composite)
An engineer was asked to approve a multi-year vendor contract with partial test results. She did not move fast. She asked for the missing tests, consulted two colleagues, and negotiated an exit clause before signing. Showing that you slow down when you cannot undo the choice is as important as showing speed elsewhere.
The rubric
| Dimension | Weak answer | Strong answer | Staff answer |
|---|---|---|---|
| Honesty about the gap | "I had no information." | Names what was known and missing. | Quantifies the gap and why waiting cost more. |
| Reasoning | "I trusted my gut." | Explains the options weighed. | Ties the choice to reversibility and initial data. |
| Risk control | Does not mention risk. | States the risk accepted. | Sets a tripwire and says what it would trigger. |
| Communication | Long setup, short action. | Action is the longest part. | Clear spoken order; ends with a lesson. |
| Result | "It worked out." | A quantified result. | A quantified result plus what changes next time. |
Three variants of this question
Tell me about a time you made a decision with limited data.
Use the same story. Lead with the gap in the data, then the door, the check and the result. If the interviewer stresses data, name which figure you wanted and what you used instead.
Describe a time you had to decide quickly without asking anyone.
Keep the gap and the door, and add who you would normally have consulted and why that was not possible. Then show how you informed them afterwards, since speed without communication reads as a risk.
What if the decision had been irreversible?
Say you would slow down. Bezos describes one-way doors as decisions to make methodically, with consultation, and his footnote notes that companies which use the light process on one-way doors go extinct before they get large. Then point to a time you did exactly that.
Common mistakes
- "I had no information, so I went with my gut." (wrong) It sounds like guessing. (right) "I had three of five signals, I knew which two were missing, and I chose on those."
- "I always move fast." (wrong) It ignores one-way doors. (right) "I move fast on reversible choices and slow down when I cannot undo them."
- "It worked because I was right." (wrong) It hides the uncertainty. (right) "I was not sure, so I set a check, and it confirmed the choice."
- Spending most of the answer on the Situation. (wrong) Both Northwestern and MIT say Action should be the longest part. (right) Keep the setup to a few sentences and spend the time on how you decided.
- Saying "we" throughout. (wrong) The interviewer cannot see your part. (right) "I decided, I set the check, and the team ran it."
Follow-up questions
What if the decision turned out wrong?
Lead with the recovery. Say how the check showed the error, what you reversed, and what it cost. Bezos writes that if you are good at course correcting, "being wrong may be less costly than you think, whereas being slow is going to be expensive for sure", and a wrong choice told honestly can be as strong as a right one. Owning a real mistake is also the core of the greatest weakness answer.
How much information is enough before you decide?
It depends on reversibility. Bezos's letter gives a hedged rule of thumb of about 70% for most decisions and calls waiting for 90% probably slow. For one-way doors, ask for more, consult, and slow down.
Is bias for action the same as acting without thinking?
No. Amazon's wording is "Many decisions and actions are reversible and do not need extensive study" together with "calculated risk taking", and AWS guidance pairs it with Dive Deep and initial data.
What if you cannot think of a good example?
Pick a smaller one: a choice on a deadline where one fact was missing. MIT suggests preparing three to five stories so one fits. The common interview questions page lists neighboring questions, and why should we hire you shows how to connect a story to the role.
- Name the gap in what you knew before you name the choice.
- Judge honestly whether the decision was a two-way or a one-way door.
- Set a tripwire that would show you were wrong, and say what you would do.
- Spend the longest part of the answer on how you decided, then give a quantified result.
- Treat the 70% figure as a hedged heuristic, not a precise rule.
Sources
- 2016 Letter to Shareholders (High-Velocity Decision Making) — Amazon.com, Inc. (Jeff Bezos), 12 Apr 2017
- 2015 Letter to Shareholders (Type 1 and Type 2 decisions) — Amazon.com, Inc. (Jeff Bezos), 1 Apr 2016
- Amazon Leadership Principles — About Amazon, 26 Oct 2026
- Leading and Innovating with Leadership Principles — Amazon Web Services Executive Insights, 26 Oct 2026
- The STAR Approach — Northwestern University Career Advancement, 26 Oct 2026
- Using the STAR method — MIT Career Advising and Professional Development, 26 Oct 2026
