A behavioural round is a structured evidence hunt. The interviewer asks for one past event, writes down what you did, and scores it against a short list of signals the company has trained them on. This chapter shows what those signals are, how the notes turn into a hire decision at Amazon, Google, Meta and Microsoft, and what separates a weak answer from a strong one when the facts are the same. By the end you should be able to score your own stories the way an interviewer would.
What the round measures
Behavioural questions predict future behaviour from past behaviour. That is the whole theory, and it explains the rules: interviewers want a specific past event (not what you would do), your own actions (not the team's), and evidence they can write down (numbers, names of systems, dates, what people said). Every company names its signals differently, but the same eight show up in almost every rubric.
Google also asks hypothetical questions ("How would you approach this situation…?"), and its re:Work guide treats them as a separate question type, used to see how you handle a situation you have not met. For those, "what you would do" is the question: answer the scenario (clarify, state assumptions, give options with their costs, decide, say how you would measure success) and then add a short real example. Everything else in this chapter is about the behavioural kind.
| Signal | What they listen for | Typical prompt | Evidence that scores |
|---|---|---|---|
| Scope | How far your decisions reached: a function, a service, a team, several teams, the org | "Tell me about the most complex project you led." | Number of teams or services affected, users or money at stake, how long the effect lasted |
| Ownership | You acted without being told, finished the job, and owned the failures too | "Tell me about something you did that was outside your job." | "I noticed, I decided, I followed up three weeks later" |
| Conflict | You raised disagreement directly, with data, and kept the relationship | "Tell me about a disagreement with a peer or manager." | The other side's view stated fairly, what you said, how it was settled, what you did after |
| Ambiguity | You made progress with missing information and said what you assumed | "Tell me about a time the requirements were unclear." | The assumption you wrote down, the cheap experiment you ran, the point where you revisited it |
| Growth | You learn from feedback and failure, and change behaviour, not only opinion | "Tell me about feedback that was hard to hear." | A before and after: what you do differently now, with a later example |
| Collaboration | You made other people's work easier and shared credit | "Tell me about working with a difficult team." | What you changed in how the teams worked, not only what you built |
| Leadership | You set direction and got others to follow without relying on a title | "Tell me about a time you led without authority." | The proposal, who you persuaded and how, who did the work, what stuck |
| Judgment | You weighed options, chose with reasons, and knew what not to do | "Tell me about a hard trade-off." | Two or three options with the cost of each, the one you picked, and what would have changed your mind |
Almost any behavioural question is two questions at once: the stated one ("a disagreement") and the level one ("how big was the thing you disagreed about?"). The same conflict story scores Senior or Staff depending on what was at stake and who you had to move.
How scoring works
The mechanics are similar at every large company. Interviewers take near-verbatim notes, map each piece of evidence to a competency, give a rating, and write a summary before they see anyone else's view. A separate meeting or committee then reads all the write-ups together. The practical consequence: you are scored on what the interviewer could write down. A clever answer with no quotable facts scores lower than a plain answer with three numbers.
question ──► your answer ──► interviewer notes (near verbatim)
│
▼
evidence mapped to competencies
(e.g. Ownership +, Conflict ?, Scope = team)
│
▼
rating + written summary, submitted
before reading other interviewers
│
▼
debrief / hiring committee reads the whole packet
│
┌──────────────┴──────────────┐
▼ ▼
hire / no hire level decision
(often set by behavioural
and design evidence)
What differs is who decides and how much the behavioural round weighs. In general terms, as reported in 2025–2026:
Amazon
Behavioural questions run through every round, not one. Amazon's own SDE III page says each interviewer is assigned Leadership Principles to evaluate, and the 60-minute phone screen is half Leadership Principle questions. Interviewers submit written feedback; the Bar Raiser, an interviewer from outside the hiring team, runs the debrief and can block the hire. The official page promises an outcome within five business days of the loop. Amazon says each interviewer typically asks two or three behavioural questions; expect 10–15 minutes of follow-up on a single answer (reported).
One round is "Googleyness and Leadership" (G&L); in India's 2026 split loop it usually comes in the first, video stage (reported). Google's re:Work guide (updated March 2026) describes structured interviews scored against rubrics with four levels (poor, borderline, solid, outstanding) and three attribute categories: role-related knowledge, problem solving, and leadership, which for individual contributors means "emergent leadership". The four-attribute list with "general cognitive ability" is the older framing, used by Laszlo Bock in 2015 and by Google's earlier re:Work guide. Prep guides list six Googleyness attributes (thrives in ambiguity, values feedback, effectively challenges the status quo, puts the user first, does the right thing, cares about the team), and Hello Interview reports G&L carries most of the L5 level decision. G&L mixes behavioural and hypothetical questions. Interviewers write feedback; a hiring committee that did not interview you reads the packet and decides, so your answer has to survive being read cold. A May 2026 pilot in some orgs adds a technical design discussion of a past project (Business Insider).
Meta
One 45-minute behavioural round, usually with a senior engineer or manager. Guides describe five scored areas: resolving conflict, driving results, embracing ambiguity, growing continuously and communicating effectively. The behavioural round carries a lot of the levelling weight: E5 vs E6 is decided mostly by design and behavioural evidence of cross-team leadership, and a weak behavioural round is reported to drop an E5 candidate to E4.
Microsoft
Behavioural questions appear in most rounds, often in the first ten minutes of a technical one. Senior loops usually end with an "As Appropriate" (AA) interview with a senior leader who reads the earlier feedback and has final approval. Guides say interviewers listen hardest for growth mindset: what you learned and what you changed.
"I did well in the coding rounds, so the behavioural round is a formality." At all four companies the behavioural evidence sets or caps your level. At Amazon a Bar Raiser concern on Leadership Principles can end a loop where every coding round passed.