Skip to content

How behavioral interviews work

For interns, new grads, and engineers with 0 to 3 years of experience. You leave with the grading rubric, a timed STAR skeleton, and a 2-week plan that ends with 8 to 10 rehearsed stories.

Why this round decides offers

  • Steve Huynh, an ex-Amazon Principal Engineer and Bar Raiser, sat on nearly a thousand interview loops. He writes that candidates who missed offers "seldom failed because they lacked technical skill." His fix: if you plan 80 hours of prep, move 10 of them to stories (Pragmatic Engineer).
  • Hello Interview reports that most down-leveling decisions come from the behavioral round, not from technical gaps (Hello Interview).
  • At Amazon, about half of an SDE loop is Leadership Principle questions (About Amazon). Each interviewer typically asks 2 or 3 of them (Amazon SDE II prep).
  • When I was prepping for final rounds at Amazon and Meta, I asked friends at Google, Amazon, and Meta where strong candidates fail. The answer was the same: the behavioral round, not coding (Why Smart Candidates Still Fail FAANG Interviews).

Do this now: put 10 to 12 hours of story prep on your calendar over the next 2 weeks, then follow the 5-step prep plan.

How the round is scored

Interviewers walk in with a fixed list of competencies and a rubric. They write notes as plus and minus signals against each competency (Hello Interview). Google scores answers on a written scale of poor, borderline, solid, and outstanding, and uses the same rubric for every candidate (Google re:Work).

What the interviewer decides

Decision What drives it What you control
Hire or no hire Positive signal on nearly every competency tested Have a story for every competency on the list
Level Scope and difficulty of your stories Lead with your biggest-scope stories
Credibility: do I believe them? Specific details, numbers, honest mistakes Use real stories with real numbers
Fit: would I work with them? How you talk about others Credit people, never blame, show you listened
Communication: can they structure thoughts? Clear order, short answers STAR, 2 to 3 minutes, answer the question asked

Sources: Hello Interview (the three implicit checks), Tech Interview Handbook rubrics (hire needs positive signal on nearly all areas for juniors).

One interviewer does not decide alone: a debrief or committee reviews all the notes (Tech Interview Handbook). At Google that is the hiring committee, and at Amazon a Bar Raiser runs the debrief with the hiring manager.

The 8 competencies most rubrics test

From the Tech Interview Handbook rubric by Lior Neu-ner, an ex-Meta engineering manager. A typical interview asks 5 or 6 questions and goes deep on each.

Competency Typical question A strong junior answer shows
Motivation Why this company? Why this team? Specific reasons tied to the product and your past work
Proactivity Tell me about something you did that nobody assigned. You spotted a problem and acted without being asked
Working in an unstructured environment Tell me about a project with unclear requirements. You asked questions, wrote a plan, and checked in early
Perseverance What did you stick with for months? You kept going through setbacks and changed approach when needed
Conflict resolution Tell me about a disagreement with a teammate. You listened, used data, reached a decision, kept the relationship
Empathy Tell me about helping a struggling teammate. You understood their view and helped without taking over
Growth Tell me about critical feedback you received. You changed a behavior and can show the later result
Communication Explain a technical topic to a non-engineer. Short, clear, adjusted to the listener

Do this: copy the 8 competencies into a doc and write one story idea next to each. Any row you leave blank is the first gap to fill in Build your story bank.

What junior scope sounds like

Level Scope of impact in your stories Source
Junior (intern, new grad, early career) Your own work or your team's focus area, usually just you TIH rubrics
Senior The whole team, three or more people same
Staff The org, two or more teams same

Hello Interview's example of a junior-level conflict is a code review debate about how to structure a class. Small scope is fine. Vague is not.

Then score each story on Huynh's four dimensions before you rehearse it (Pragmatic Engineer):

Dimension Ask of your story Entry-level target
Scope How many people did my work affect? Me, and it starts to help teammates
Contribution What did I do, not the team? Every action sentence starts with "I"
Impact What changed, in numbers? Time saved, bugs prevented, users, latency, grade
Difficulty What made it hard? A real constraint, trade-off, or obstacle

The four question types

Type Example How to answer
Past behavior Tell me about a time you missed a deadline. One real story in STAR, 2 to 3 minutes
Open-ended Walk me through the project you are most proud of. Offer 2 or 3 options, let them pick, then STAR
Philosophy How do you approach code reviews? Your principle in one line, then one short real example
Hypothetical What would you do if a teammate took credit for your work? Clarify, state assumptions, compare options, choose, tie to a real example

The first three types come from Hello Interview. Hypotheticals are common at Google; see Googleyness and Leadership.

STAR, timed

STAR means Situation, Task, Action, Result. Amazon recommends it and also says it is not the only valid structure (Amazon interview loop). Google's tips page lists the same four parts, with "Activity" in place of Action (Google interview tips).

Amazon says a STAR answer typically runs 2 to 3 minutes (About Amazon). MIT's career office splits the time 20/10/60/10 (MIT CAPD). Applied to a 2.5-minute answer:

Part Share Seconds What to say
Situation 20% 30 Where, when, your role, the stakes, one number
Task 10% 15 What you owned and how success was measured
Action 60% 90 3 to 5 steps you took, each with a reason, all with "I"
Result 10% 15 The number before and after, and who benefited
Learning (optional) extra 10 to 15 One lesson and where you used it since

Copy this skeleton for every story:

Situation (30 s): In [month year], I was [role] on [team or project] at [place].
  [The problem, with one number]. [What happens if nobody fixes it].
Task (15 s): I owned [specific piece]. Success meant [measurable goal] by [date].
Action (90 s): First, I [step] because [reason]. Then I [step].
  The hard part was [obstacle], so I [what you did].
  I [worked with / convinced] [person or team] by [data or argument].
Result (15 s): [Metric] went from [before] to [after] in [time]. [Who benefited].
Learning (10 to 15 s): I learned [one lesson]. Since then I [habit],
  for example on [later project].

Variants: pick one and use it every time

Framework Parts When it helps Source
STAR Situation, Task, Action, Result Default for every company Amazon, MIT CAPD
STAR-L STAR plus Learning Failure and feedback questions Rutgers
STARI STAR plus Improvement Amazon; a Bar Raiser recommends ending with what you would improve About Amazon
CARL Context, Action, Result, Learning When your Situation and Task blur together University of Edinburgh, taught by Hello Interview
SPSIL Situation, Problem, Solution, Impact, Lesson Same idea, IGotAnOffer's version IGotAnOffer

Five rules for every answer

  1. Say "I" for what you did. Amazon's loop page asks you to focus on your individual contribution and say "I" instead of "we" (Amazon). Keep "we" for the team's outcome.
    • Jugal's version: "'We built this feature' means nothing. 'I designed the caching layer, reducing latency by 40%', that's what they want to hear." (Your 6-Week Amazon Interview Roadmap)
  2. Put a number in the result. Use Google's formula: accomplished [X], as measured by [Y], by doing [Z] (Google interview tips). If you have no exact number, give an honest estimate and say "roughly" (interviewing.io), or use the ideas in metrics when you have none.
  3. Keep the setup under 30 seconds. Time it (IGotAnOffer).
  4. Answer the question asked. A failure question needs a failure. A conflict question needs two people who disagreed.
  5. Stop and check in. Amazon suggests pausing after your answer and asking whether the interviewer wants more context (Amazon).

The 5-step prep plan

Budget 10 to 12 hours over 2 weeks. That matches Huynh's suggestion of about 10 hours of story prep.

  1. Decode what your company scores (1 hour). Open your company's page in the table below and the job description. Write down the 6 to 8 values or competencies you will be graded on. For Amazon, a Bar Raiser says the job description usually shows which principles matter most (About Amazon).
  2. Collect 20 to 30 moments and keep 8 to 10 (3 hours). Follow Build your story bank. It uses interviewing.io's selection method: brainstorm, scope, add evidence, filter, title (interviewing.io).
  3. Write each story as a STAR card with a number (3 hours). Use the story card. Map each story to 2 or 3 company values. Then do an "I" audit: replace every "we" that hides your own action.
  4. Rehearse out loud and record (2 hours). Time each story, then have a friend ask "why?" and "what exactly did you do?" five times per story.
    • Jugal's rule: "Record yourself. Watch it back. Cringe. Do it again. The first time you tell these stories, they'll sound scripted. By the tenth time, they'll sound natural." (Your 6-Week Amazon Interview Roadmap)
  5. Run 2 or 3 mocks and fix the weakest story (2 to 3 hours). Book a free peer on Aced Practice, do one AI session on Free Mock Interview, or pay for one on interviewing.io. Ask your partner to grade each answer as poor, borderline, solid, or outstanding. More options: Mock interviews.
    • After each mock, ask two of Jugal's review questions: "Did I talk enough?" and "Would you hire me based on this interview?"

Two-week schedule

Days Do this Output
1 Step 1: list what your company scores 6 to 8 competencies on one page
2 to 3 Brainstorm moments from every source 20 to 30 one-line moments
4 to 5 Score, filter, write 5 story cards 5 STAR cards with numbers
6 to 7 Write 3 to 5 more cards, fill the mapping grid 8 to 10 cards, grid with no empty rows
8 to 10 Rehearse 3 stories a day out loud, recorded Each story at 2 to 3 minutes
11 to 12 Two mocks, one peer and one AI or paid Notes on your 2 weakest answers
13 Rewrite the 2 weakest stories, re-record Fixed stories
14 Prepare "Tell me about yourself", "Why [company]", and questions to ask Three short scripts
  • I know the 6 to 8 competencies my target company scores.
  • I have 8 to 10 stories, each with at least one number.
  • Every row of my mapping grid has at least 2 stories.
  • I have a real failure story with a lesson I applied later.
  • Each story runs 2 to 3 minutes out loud.
  • I recorded myself at least once per story.
  • I did 2 mocks and fixed what they exposed.
  • I have a 60 to 90 second "Tell me about yourself".
  • I have 3 questions ready for each interviewer.

Red flags that sink answers

Red flag What it sounds like Fix Source
Blaming others "The backend team kept breaking things." Describe what you did and what you learned About Amazon, Hello Interview
Negative talk about an employer "My last manager had no idea what he was doing." Talk about the situation, not the person About Amazon
Vague or rambling Five minutes of context, then "so yeah, it worked out" 30-second setup, numbers, a clear end Pragmatic Engineer
No numbers "It made things a lot faster." "From 40 minutes to 24 minutes" interviewing.io
"We" everywhere "We designed, we built, we shipped." "I" for your actions, "we" for the outcome Amazon
No stakes The interviewer can't tell why it mattered One line on what happens if nobody acts Hello Interview
Too low-level too fast Function names in the first 20 seconds Start with the problem and the people, then go technical when asked same
Scripted delivery Word-perfect, flat, no pauses Memorize bullets, not sentences Amazon worksheet
Asking "is that good enough?" after every answer Seeking approval instead of offering detail Ask once or twice, as an offer to go deeper Hello Interview
Arguing who was right A conflict story that proves the other person wrong Focus on how you reached a decision together interviewing.io
Solo hero "I carried the whole team." Show how you worked with others Hello Interview Google L4
Fake failure "I work too hard." A real miss with consequences and a lesson IGotAnOffer
Confidential details Unreleased features, revenue numbers Use public information or relative numbers About Amazon
Invented stories A story built to fit the question Use a real story or say you have not faced it interviewing.io

Do this: play back each recorded story and tick off every row above it hits. Fix those lines before the next take.

Watch out: interviewing.io's advice on a missing story is that no answer is often better than a bad one. Use the bridge script below instead of inventing a story.

Scripts for hard moments

Mind goes blank (from Jugal's Amazon roadmap):
"Give me 30 seconds to think through this."

No story fits exactly (bridge; idea from Hello Interview):
"I haven't faced that exact situation. The closest was [situation],
where I [action]. Want me to walk through that one?"

Open-ended question (offer a menu; idea from Hello Interview):
"I could talk about [a technical trade-off], [a disagreement on my team],
or [a production bug I owned]. Which is most useful for you?"

Mid-story fork:
"From here I can go into the technical details or how I got the team
to agree. Which would help more?"

Lost the thread:
"Let me restate that more clearly. The short version is [one sentence]."

After a long answer (once or twice per interview; idea from Amazon's loop page):
"Is that the level of detail you want, or should I go deeper on any part?"

Tell me about yourself

The Tech Interview Handbook lists it among the three questions to prepare first, and its self-introduction guide says it is almost always asked first (Tech Interview Handbook). Keep it to 60 to 90 seconds. Jugal's 6-step pitch (hook, introduce yourself, highlight value, align with the audience, call to action, practice) works here with small changes (Craft the Elevator Pitch).

Present (20 s): I'm [name], a [year and degree] at [school] / [role] at [company].
  Most of my work is in [area].
Proof (30 s): Recently I [project or internship result with a number].
  Before that, I [second proof with a number].
Why here (20 s): I'm interested in [team or company] because [specific product,
  problem, or technology you have actually used or built with].
Hand-off (5 s): Happy to go deeper on any of those.

Write it, read it aloud with a timer, and cut until it fits 90 seconds. For the 30-second career fair version, see Career fairs.

Questions to ask at the end

Prepare 3 per interviewer. Do not repeat the same question across a loop. Pick from the list below or from the Tech Interview Handbook's final questions.

  • What does success look like in this role in the first 90 days? (Grow with Google)
  • What are the biggest challenges the team faces now, and how would this role help? (same source)
  • What is the code review and design review process like? (awesome-behavioral-interviews)
  • What does onboarding look like for a new grad here? (same source)
  • What is the most interesting project you have worked on here? (same source)
  • At Amazon, ask about users: interviewing.io suggests framing your question around the customer experience to show Customer Obsession (interviewing.io).

Use AI to prepare, never during the interview

Google states that AI tools are not permitted during its interviews (Google: how we hire). Use AI before, as a skeptical practice partner. Jugal's rule: "AI gives you a 70% draft fast, your job is the last 30% that makes it true and yours." (The job-search tool stack I'd actually use in 2026)

  1. Generate questions for your exact role. Jugal's starting prompt is "Give me 5 behavioral questions for a [role] role at [company]" (Jugal's AI job application guide). Add each question to your grid.
  2. Answer one out loud, paste your answer into the prompt below, and fix what it flags.
You are a [Company] interviewer scoring a behavioral answer from a [intern / new grad]
software engineer. Here is my story in STAR form:
[paste story]

1. Ask me 5 follow-up questions, one at a time, like a skeptical interviewer.
2. After my answers, grade the story as poor, borderline, solid, or outstanding,
   and say why.
3. List every sentence where I said "we" but should say "I".
4. List every claim that needs a number.
5. Do not invent details or numbers. If something is missing, ask me for it.

Pick your company page

Company Page What dominates
Amazon Amazon Leadership Principles 16 principles, the Bar Raiser, about half the loop
Google Googleyness and Leadership One G&L round, hypotheticals, hiring committee, team match
Meta Meta Execution, ownership, and speed
Microsoft Microsoft Company values and the as-appropriate interviewer
Apple Apple Collaboration, craft, product thinking
Netflix Netflix The culture memo and the keeper test
Others Other companies Uber, Airbnb, Stripe, Bloomberg, Databricks, Nvidia, quant firms and more

Each company also has a process page under Companies.

Resources

Free first.

From Jugal's Substack:

Next: Build your story bank