Google's process is unusual in one specific way: the people who interview you do not decide whether you are hired. They gather evidence, write it up, and pass it to a hiring committee that never meets you. Everything else about preparing for Google follows from that fact — your job in each round is to produce written evidence against four named attributes, clearly enough that someone reading a document three weeks later reaches the right conclusion. This guide covers the full pipeline, the attributes you are scored on, what each round tests, and why the timeline runs so long.
The Full Pipeline, Stage by Stage
Application or referral
A referral gets your CV read by a human faster; it does not lower the bar. Google is explicit that referrals affect visibility rather than standards, and the screening criteria are unchanged.
Recruiter screen — 30 minutes
Background, target level, timeline, and logistics. Your recruiter is also your best source of process information — ask which rounds you will have, whether a system design round is included, and what level the role is targeting. All three answers change how you prepare.
Technical phone screen — 45 minutes, sometimes two
One or two coding problems in a shared editor. Medium difficulty, and the bar is a working solution with correct complexity analysis, arrived at while thinking out loud. Some roles substitute an online assessment at this stage.
The onsite loop — 4 to 5 rounds of 45 minutes
Two or three coding rounds, a system design round at L5 and above, and one behavioral round. Often run virtually. Interviewers deliberately do not compare notes before writing their feedback, so each round starts from zero.
Feedback packet assembled
Each interviewer submits scores against their assigned attributes plus written evidence, commonly including your code near-verbatim. This packet is what gets decided on.
Hiring committee review
A committee of Googlers uninvolved in your loop reads the packet and votes on hire, no hire, or a level adjustment. They may send it back for additional information, which adds a week or more.
Team matching
You talk to hiring managers with open roles until there is a mutual fit. Passing the bar with no available match means waiting — this stage causes most of the unexplained late-stage silence candidates experience.
Final approvals and offer
Senior review, then compensation. Six to ten weeks total is normal; longer happens, particularly at senior levels or during a hiring slowdown.
Why the process feels slow, and what it does not mean
The committee structure exists to reduce individual interviewer bias, and it costs calendar time. A two-week gap after your onsite is usually queue time or team matching, not deliberation about you specifically. Ask your recruiter for a status update every seven to ten days — it is expected and it will not count against you.
The Four Attributes You Are Scored Against
Every interviewer is assigned specific attributes to assess. Knowing which behaviours map to which attribute lets you supply evidence deliberately instead of hoping it comes across.
| Attribute | What it means | How to give evidence for it |
|---|---|---|
| General Cognitive Ability | How you reason through problems you have not seen before | Narrate your approach, state assumptions, work through a wrong idea out loud rather than silently discarding it |
| Role-Related Knowledge | Depth in the craft the role actually requires | Correct complexity analysis, sound data-structure choices, specifics from your own past work |
| Leadership | Influence and initiative without formal authority | Stories where you drove something nobody assigned you, or changed a decision by persuasion |
| Googleyness | Comfort with ambiguity, collaboration, intellectual humility | Ask clarifying questions before designing; concede quickly and correctly when an interviewer finds a flaw |
Note what the fourth attribute rewards. Intellectual humility is scored, which means conceding a point well is worth marks rather than costing them. Candidates who defend a broken approach for five minutes lose on Googleyness even when they eventually fix the code.
The Coding Rounds
Google coding rounds are conventional data structures and algorithms, delivered in 45 minutes with the expectation that you talk continuously. The topic distribution has been stable for years:
Very common
- Arrays, strings, two pointers
- Hash maps and sets
- Trees and graph traversal
- Binary search, including on answers
- Sorting and interval problems
Common
- Dynamic programming, usually one dimension
- Heaps and top-k problems
- Backtracking
- Sliding window
- Stacks and monotonic stacks
Rare
- Advanced graph algorithms
- Segment trees, tries beyond the basics
- Heavy mathematical puzzles
- Brainteasers — dropped years ago
What actually decides these rounds, in order of weight:
- Clarifying before coding. Input size, duplicates, empty input, whether the input can be mutated. Thirty seconds here scores on two attributes at once.
- A stated brute force, then an improvement. Naming the naive solution and its complexity before optimising shows the reasoning path. Jumping straight to a memorised optimal solution shows only that you have seen the problem.
- Continuous narration. Interviewers write down what you say. Silent thinking produces an empty feedback packet, and the committee cannot score what was never recorded.
- Correct complexity analysis. Time and space, said explicitly and correctly. Getting this wrong after a working solution damages the Role-Related Knowledge score disproportionately.
- Testing your own code. Walk one real input through it, including an edge case, before declaring you are done. Finding your own bug is a positive signal; having the interviewer find it is not.
System Design and the L4 / L5 Line
System design is included from L5 upward and it is usually the round that decides your level. Two candidates with identical coding performance are separated here: the one who scopes requirements, estimates scale, justifies each component, and names their own bottleneck lands L5, while the one who produces a correct but unexamined diagram lands L4.
The structure that works is the standard five-phase approach — clarify requirements, estimate scale, define the API and data model, draw the architecture, then stress it. Our system design framework covers each phase with the time budget and the twelve prompts that recur, and it applies directly to Google's version of the round.
Level is negotiable earlier than you think
Ask your recruiter at the screen which level the loop is targeting. If it is below where you believe you sit, that is the moment to make the case, because the loop is calibrated to a level before it starts. Arguing for a level after the hiring committee has voted is far harder than setting expectations at the beginning.
The Behavioral Round: Googleyness and Leadership
Standard STAR format, with Google weighting influence-without-authority more heavily than most companies. The recurring themes are consistent enough to prepare directly against:
Driving something nobody assigned you
The core Leadership signal at Google. A project you identified, scoped, and got others to support without being the manager. Be specific about how you got buy-in — that mechanism is the part being assessed.
Working through ambiguity
A problem where the goal itself was unclear. Show how you narrowed it: the questions you asked, the assumptions you wrote down, the smallest thing you shipped to learn something. This is the Googleyness attribute in its purest form.
Disagreeing with someone senior
How you made the case, and what you did once the decision went against you. Committing fully after losing the argument is the part interviewers listen for.
A failure you own
Something that genuinely went wrong, with your contribution stated plainly and the resulting change described. Deflected blame reads as a humility failure, which is directly scored here.
Cross-functional collaboration
Working with product, design, or another engineering org, ideally where incentives conflicted. Google runs large matrixed teams and this is a daily reality of the job.
Eight to ten well-prepared stories cover this comfortably, since one story can serve several themes depending on emphasis. Our behavioral interview guide covers the STAR structure in depth, and it is worth contrasting Google's approach with Amazon's Leadership Principles loop if you are interviewing at both — the formats look similar and the emphasis is genuinely different.
Google vs Amazon: Where the Preparation Differs
| Amazon | ||
|---|---|---|
| Who decides | Hiring committee, never meets you | Loop debrief, Bar Raiser holds a veto |
| Rubric | Four broad attributes | 16 named Leadership Principles |
| Behavioral weight | One dedicated round | Scored in every round, technical ones included |
| Coding emphasis | Algorithmic depth and reasoning | Practical data structures, less exotic |
| Follow-up style | Probes your approach mid-problem | Three to five drill-downs per story |
| Timeline | Six to ten weeks, team matching adds time | Two to six weeks, single-day loop |
A Realistic Preparation Timeline
Weeks 1–3: algorithm fluency
Work the very common column above until the patterns are recognition rather than recall. Volume matters less than reviewing every problem you failed and being able to explain why the right approach was right.
Weeks 4–5: system design, if you are targeting L5 or above
Twelve prompts, out loud, timed to 45 minutes, using the five-phase structure. Repeat three of them a second time to build pace.
Week 6: behavioral stories
Write eight to ten stories in STAR form, weighted toward influence without authority and ambiguity. Rehearse them spoken and timed to two minutes.
Week 7: full mock loops
Two or three complete loops back to back, in one sitting, with narration throughout. Interview stamina is a real and separately trainable constraint — four consecutive 45-minute rounds is genuinely tiring.
Final days: taper
Light pattern review, confirm your logistics, and sleep. Cramming the night before measurably degrades exactly the reasoning-out-loud ability the loop is designed to measure.
Notice how much of that plan is spoken rather than written. Google's process scores what you say while you work, so silent practice trains the wrong skill entirely. This is where unlimited AI practice is genuinely useful — the same problem, narrated, as many times as you need, at any hour. We have compared AI and human coaching honestly, including where a human still wins, and our FAANG preparation roadmap covers sequencing all three round types without burning out.
Practise narrating your reasoning, not just solving
Amigo runs unlimited timed practice from your resume and target role, then supports you live during the real interview with structured answers streamed in real time.
Try Amigo free →Frequently Asked Questions
How long does Google's interview process take?
Commonly six to ten weeks end to end, and longer is not unusual. The interviews themselves take two to three weeks; the remaining time goes to hiring committee review, team matching, and final approvals. Long gaps are structural rather than a signal about your performance.
What are Google's four evaluation attributes?
General Cognitive Ability — how you reason through unfamiliar problems; Role-Related Knowledge — depth in the craft the job needs; Leadership — influence and initiative without formal authority; and Googleyness, meaning comfort with ambiguity, collaboration, and intellectual humility. Every interviewer scores against specific attributes and submits written evidence.
Who actually decides whether Google hires you?
A hiring committee of Googlers not involved in your interviews. They read the written feedback packet — your interviewers' notes, scores, and often your transcribed code — rather than meeting you. This is why what your interviewer writes down matters as much as the impression you made.
What is team matching at Google?
After you pass the interview bar, Google matches you with a team that has an open role, usually via short conversations with hiring managers. Passing the bar without an available match means waiting, and it is the most common cause of a long silence late in the process.
Does Google still ask brainteasers?
No. Google publicly dropped brainteasers years ago after concluding they did not predict job performance. Current rounds are data structures and algorithms, system design at senior levels, and structured behavioral questions.
What level will I be hired at?
For software engineering, L3 is entry level, L4 is roughly two to five years of experience, L5 is senior, and L6 is staff. Level is set by how your interviews go, not only by your years of experience, and system design performance is usually what separates L4 from L5.
How many rounds are in a Google onsite?
Typically four to five: two or three coding rounds, a system design round for L5 and above, and one behavioral round on Googleyness and Leadership. Rounds are usually 45 minutes and are frequently virtual even when described as onsite.
Can you reapply to Google after a rejection?
Yes. The usual guidance is to wait six to twelve months, and your prior feedback packet is retained and visible to future recruiters. Candidates are hired on a second or third attempt regularly, so a rejection is genuinely not permanent.
Found this useful?
Share it with someone preparing for an interview.
