Most Microsoft interview guides spend their length on a round that Microsoft does not officially name. Search for Microsoft interview prep and you will find pages of advice about the "AA," the As Appropriate interviewer, usually described as a gatekeeper with veto power who appears a week or two after your loop if you pass. Meanwhile Microsoft publishes six named competencies it evaluates candidates against, on a public page, and almost none of those guides mention them.
That gap is the most useful thing to know about preparing for Microsoft. Candidates are optimizing against folklore while the company's actual rubric sits in the open.
This guide walks the process stage by stage using what Microsoft publishes, plus the few outside sources that hold up. Where something is widely repeated but unsourced, it says so. There is a lot of that in this topic.
What Microsoft actually publishes
Start with the rubric, because it is the part you can prepare against directly.
Microsoft's interview tips page lists six core competencies. Collaboration, meaning effective communication in a team. Drive for results, meaning delivering on commitments and holding yourself accountable. Customer focus. Influencing for impact, meaning persuasion through communication. Judgment, meaning problem scoping and business acumen. Adaptability, meaning handling ambiguity with agility.
Read that list again and notice how much of it is not technical. Four of the six are about how you operate with other people and how you make calls under uncertainty. If your prep consists entirely of algorithm practice, you are preparing for one competency out of six.
For technical roles, Microsoft's technical interviewing page names four areas pulled from the engineering lifecycle. Problem solving, which it specifies happens inside a 45 minute round. Design. Coding, described as clean, concise, and bug free in your strongest language. Testing, explicitly including corner cases and security implications.
Testing is the one candidates skip. It is listed as a named evaluation area, and most people treat it as a courtesy sentence at the end of a coding round. Walking through corner cases unprompted is a cheap way to hit something the rubric asks for.
On values, Microsoft's how we hire page names respect, integrity, accountability, and growth mindset, and says it evaluates how you will learn, collaborate, and contribute across teams. The interview tips page puts it more bluntly, saying Microsoft values "true learners" and wants "learn-it-alls instead of know-it-alls," which is Satya Nadella's phrasing carried into hiring copy. Growth mindset is genuinely part of the stated criteria rather than wall art. What does not exist, despite what some prep sites claim, is a published growth mindset scoring band. There is no rubric with numbered levels. Treat it as a disposition the interviewers are listening for, not a scale you can game.
The shape of the Microsoft interview process
Microsoft describes seven stages from job search to hire, and the part you control runs from application review through interview to offer.
On loop size, Microsoft says most interviews include two to four conversations with potential teammates and cross-functional colleagues, each up to an hour. Nearly every prep site says four to five. The honest answer is that Microsoft's number is the only sourced one, and candidate reports skew toward larger loops, probably because the people who write up their experiences online are disproportionately interviewing for senior engineering roles at large orgs. Plan for up to five, and do not treat a three-round loop as a bad sign.
Format is worth flagging because the internet is confidently wrong in both directions. Microsoft's current careers pages say interviews happen over the phone, on Microsoft Teams, or in person. Not "entirely virtual," which older prep content still quotes, and not a wholesale return to onsite loops either. Microsoft did move employees to three days a week in the office, with anyone living within 50 miles of an office expected onsite on that schedule, rolling out to Puget Sound first by the end of February 2026, per chief people officer Amy Coleman's September 2025 update. Whether that pushed interview loops back in person is something no source establishes. Ask your recruiter rather than guessing.
One more structural fact. Microsoft generally does not give individualized feedback to candidates who are not selected. If you want to know why you did not advance, the recruiter conversation during the process is where you gather signal, because there will not be a debrief afterward.
The recruiter screen
The recruiter screen is short and does two things. It confirms your background plausibly maps to the role at the level it is posted at, and it produces the summary of you that travels through the rest of the process.
That second function is the one candidates underweight. Your interviewers read something before they meet you. If you cannot describe your own scope crisply, someone else writes that description, and they write it worse than you would. Practice a 60 second version of what you own, what you shipped, and what the outcome was. Not a career history. A summary that survives being written down and read by a stranger.
Microsoft also states its expectations on AI use here, and it is more permissive than most companies. Candidates are encouraged to use AI tools "responsibly and ethically" while preparing, as long as the result reflects their true capabilities. During assessments and interviews, candidates are expected to demonstrate their own skills without outside assistance unless explicitly permitted. Preparation with AI is fine. Assistance during the interview is not. That line is drawn in Microsoft's own words, which is useful given how many companies leave it ambiguous.
The technical rounds
Microsoft's published topic list runs across algorithms, data structures, system design and distributed systems, AI and machine learning, data science, and networking, and it explicitly calls that list representative rather than exhaustive. The topics move with the role. The four evaluation areas do not.
The practical read on the 45 minute problem solving round is that it is not enough time to solve a hard problem silently and then explain yourself. Narrate as you go. The interviewer is writing down what they observed, and unspoken reasoning does not make it into the notes.
For design rounds, scope out loud before you draw. System design interview questions covers the shapes that come up most, and the technical interview preparation guide covers the rest of the loop. Judgment, one of the six competencies, is defined as problem scoping and business acumen. A design answer that starts by clarifying constraints and naming what is out of scope is hitting a named competency before it touches architecture.
Behavioral material tends to be woven into technical rounds rather than isolated in a single behavioral interview, though this varies and Microsoft does not publish a rule about it. Raymond Chen's description of Microsoft loops notes that each interview assesses technical competence alongside initiative, ability to learn, and teamwork, which supports the woven model without confirming that no separate behavioral round exists anywhere.
The As Appropriate round, minus the folklore
Here is what is actually documented. In 2023, Microsoft engineer Raymond Chen wrote a glossary post on the term for exactly the reason you would expect, which is that people kept encountering it without knowing what it meant.
The AA was the final interviewer in the loop, typically a senior person in the hiring group. Originally the round was conditional, hence the name. That interviewer read the earlier feedback and decided whether the candidate was still worth pursuing, either running a final interview or sending the candidate back to the recruiter without one. Chen then writes two things that most guides omit. The As Appropriate stopped being optional, so every candidate now gets that interview regardless of prior feedback. And the term is no longer officially used, though you may still hear it informally from old-timers.
So the version circulating on prep sites, where the AA is a separate interview scheduled one to two weeks later only if you clear the loop, contradicts the only Tier 1 source on the subject. The veto power framing has no primary source at all. Neither does the claim that the loop team votes and then the AA weighs in, which reads like Amazon's debrief process transplanted onto a company that does not document one.
What you should take from this is narrow and still useful. Expect the last conversation in your loop to be with someone more senior than the rest, and expect them to have read the earlier feedback. That is a real dynamic. It means the last round is the one most likely to probe a soft spot the earlier rounds surfaced. If a coding round went badly, assume the senior interviewer knows.
Why this is not Amazon's Bar Raiser
The comparison gets made constantly and it obscures more than it explains.
Amazon publicly documents the Bar Raiser program. Bar Raisers are deliberately not on the hiring team. There are more than 3,600 of them. They are nominated, trained on the Leadership Principles, and shadow other Bar Raisers for anywhere from three months to a year before running loops. They facilitate the debrief and share the hiring decision with the hiring manager.
Microsoft documents none of that. There is no published training program, no selection process, no debrief role, no headcount. The asymmetry is the point. Amazon built a formal institution to defend the hiring bar and wrote about it. Microsoft has a senior final interviewer and a piece of retired internal slang. The Amazon interview process and the Google hiring process are both worth reading alongside this one, because all three companies solve the same calibration problem differently.
That difference should change your prep. At Amazon, the Bar Raiser is a known role with a known rubric, and preparing for the Leadership Principles is preparing for that person specifically. At Microsoft, there is no equivalent secret round to crack. The six competencies are the rubric, and they apply to every conversation in the loop.
Leveling and what it means for your loop
Microsoft publishes no official level to title map. The ladder that circulates comes from community sources, primarily levels.fyi, which reports 59 and 60 as SDE, 61 and 62 as SDE II, 63 and 64 as Senior SDE, 65 through 67 as Principal, 68 and 69 as Partner, and 70 and above as Distinguished Engineer or Technical Fellow. Community-derived, consistent across sources, not company published.
Candidates report that feedback is submitted against a specific level, and that the level can move as a result of the loop, including outcomes like a hire at 61 alongside a no-hire at 60. That is a reasonable thing to plan around and not something any company source confirms, so hold it loosely.
The practical implication is that vagueness about scope costs you money. If your examples do not make clear what you owned versus what your team owned, the person writing feedback has to guess, and guesses round down.
What is overrated
Grinding algorithm problems to the exclusion of everything else. Four of Microsoft's six competencies are not technical. The loop is two to four conversations with people who will work alongside you, evaluating collaboration, influence, judgment, and adaptability.
Trying to identify and beat the AA round. It is the last interview with a senior person. There is no separate rubric published for it, and the term is retired internally.
Timeline benchmarks from prep sites. Microsoft publishes no timeline. The four to eight week figures that circulate come from aggregator content, and the more precise-sounding ones, like a 16.8 day average response, come from SEO blogs with no stated methodology. Ask your recruiter about your specific req rather than trusting a number invented to fill a page.
Reading the layoffs as a closed door. Microsoft cut 4,800 roles in July 2026, about 2.1% of its global workforce, and redeployed more than 4,000 employees internally over the prior year, per Amy Coleman. Coleman was direct that "the roles eliminated today are not being replaced by AI," while noting AI is changing how work gets done. Reqs stay open through restructuring. What changes is that internal candidates are in the pool.
A prep plan that matches the rubric
- Write one story for each of the six competencies. Collaboration, drive for results, customer focus, influencing for impact, judgment, adaptability. Use different projects so you are not retelling one story six ways.
- Practice narrating while you code. Silence costs you in a 45 minute round because your reasoning has to end up in someone's written notes.
- Build a scoping habit for design rounds. Constraints and out-of-scope first, architecture second.
- Rehearse a 60 second scope summary for the recruiter screen, with the level you are targeting stated out loud.
- Add corner cases and security implications to every coding answer, because testing is a named evaluation area.
- Run at least one full loop out loud before the real one. Reading answers is not the same as delivering them, and the gap between knowing your answer and delivering it under pressure is where most loops are lost. Voice mock interviews are built for exactly this.
Where this is heading
Microsoft's process is less mysterious than its reputation. The company publishes its competencies, names its technical evaluation areas, states its position on AI use, and declines to publish a hiring committee because there does not appear to be one. Most of the mystique attaches to a round whose own name was retired.
That pattern is worth generalizing beyond Microsoft. As AI sits on both sides of every application, prep content is getting cheaper to produce and less accurate, and folklore now outranks primary sources on most company hiring queries. The candidates who do well are the ones who read what the company actually wrote down. For Microsoft, that is one page listing six competencies, and it takes about four minutes.