By Mufy Pachorawala, founder of getAIwork · Last updated: September 19, 2026
Key facts
- The interview is conducted by an AI, on camera, and recorded. Mercor retains the video, the audio and an AI-generated transcript, and will not give you a copy of any of them
- About 20 minutes, though it varies by listing and at least one role specifies 25. An interview that is too short does not count as completed
- You get three retakes across all applications using that interview, and only your most recent attempt counts. A fourth needs an email with evidence of a technical fault, and is not promised
- There is a free practice interview, not tied to any application, never shared with companies, and you can take it as many times as you like
- Firefox and mobile browsers are not supported. Chrome is recommended; Edge and Safari work
- 618 of 1,662 live listings on the free getAIwork board are paid platform programmes like Mercor, as of September 24, 2026
What questions does Mercor ask?
Nobody outside Mercor can honestly tell you. The company publishes no sample questions, no example transcript, no question bank and no rubric. We read its entire talent help centre, all forty of its blog posts and its terms of service, and there is not one example anywhere.
It also asks candidates not to share them: “You may discuss general experiences with Mercor interviews; however, please refrain from sharing specific questions, screenshots, or any confidential information.” So a page promising “the 25 real Mercor interview questions” is either fabricated, or is somebody breaking that rule with questions that were generated for a role you are not applying to.
That is less unhelpful than it sounds, because the questions are not fixed. Mercor describes an “AI interviewer” that poses role-specific questions, generates a transcript and evaluates your performance for that listing. The same interview can also be weighted differently by different listings: “Even if two listings utilize the same assessment, the scoring and weighting may vary according to the specific requirements of the role.” Memorising answers is the wrong preparation for a system built this way. Understanding the format and the criteria is the right one, and both are published.
What the Mercor AI interview actually is
| Detail | What Mercor states |
|---|---|
| Who conducts it | An AI interviewer. There is no human alternative: it is “the standard and only method for screening candidates”, and “interviews cannot be conducted via chat, call, or any other medium” |
| Length | “Most interviews require approximately 20 minutes, although some may be shorter or longer.” One listing specifies a 25-minute conversational interview |
| Format | On camera, with voice. The glossary calls it “An automated, on-camera interview”. “Ensure your full face is visible and your voice is clear throughout” |
| Equipment | A computer with a working camera and microphone. The Start Interview button stays greyed out until the system checks pass |
| Browsers | Chrome recommended, Edge and Safari supported. Firefox and mobile browsers are not supported |
| Screen sharing | Required for some roles, and full-screen only |
| Whiteboard | Some interviews include one. It autosaves and needs no submission |
| Captions | On by default, and can be turned off |
| Recording | “We retain your video/audio recordings and an AI-generated transcript of official interviews.” You cannot get a copy: “we’re not able to share interview recordings or transcripts” |
| Minimum length | “If an interview does not satisfy the minimum completion criteria (such as duration), it will not be classified as completed” |
All quotations are from Mercor’s own talent help centre, read on September 17, 2026.
One mechanic that catches people out before they even reach the interview. Mercor runs an automated compatibility check the moment you apply, called Application Fit Guidance, with three outcomes: application allowed, a compatibility warning where you may still apply, or application blocked, where you cannot submit at all. That decision is made on your profile, before any interview happens.
What does Mercor score you on?
Four things, in its own words. Its assessments are stated to measure “your domain expertise and experience”, “your approach to solving real-world problems”, “your ability to follow instructions and manage ambiguity”, and “the quality and clarity of your written or verbal responses”.
The inputs are broader than the interview alone: “Our artificial intelligence evaluates your suitability… through the analysis of your résumé, interview performance, and publicly available profile data, such as GitHub.” If you have a GitHub account attached to the name on your CV, it is part of the assessment whether you intended it to be or not.
Mercor also states a fairness position, that “every talent undergoes the same assessment process for a given role” and that “personal characteristics such as age, gender, and ethnicity are never part of the evaluation”, and describes a limited human role: reviewers “may review assessments to verify suitability”, but their feedback “enhances the models” and is “never utilized for making hiring decisions”.
How to prepare for the Mercor interview
Everything below comes from something Mercor itself publishes. You do not need a course for this.
- Take the practice interview first. It is free, it is not attached to any application, it is never shared with companies, and you can repeat it as many times as you like. It exists precisely so that your first experience of an AI interviewer is not the one that counts. Skipping it is the most common avoidable mistake.
- Fix your CV to Mercor’s stated spec before you apply. It is parsed by AI, so the format matters: PDF only, text-based rather than a scan, one to two pages, under 2MB, a simple standard layout, no images or tables, and the headings Education, Experience, Skills and Certifications. A beautiful two-column design with an embedded photo is exactly the thing that parses badly.
- Check your equipment in daylight. Camera, microphone, Chrome, a stable connection, and a background you are happy to have recorded. The interview will not start until the checks pass, and that is not the moment to discover a driver problem.
- Answer at length. Duration is explicitly part of whether an interview counts as complete. Terse answers are a real risk here in a way they are not with a human, who can prompt you.
- Show the reasoning, not just the conclusion. Two of the four stated criteria are about approach and about clarity of expression. Walk through how you would attack a problem out loud.
- Do not use an LLM during the interview. Mercor’s policy permits grammar and wording checks, and forbids using a model “to compose your responses, rate model responses, generate rationales or explanations, or evaluate code”. You are on camera and being transcribed.
If you have done one of these before, the shape will be familiar. We covered the same territory for a different platform in how to pass the Zara AI interview, and the general principle of treating a screening as a work sample rather than an exam is the same one that runs through the DataAnnotation Starter Assessment.
The getAIwork board is free and public, screened by a human. Or take the 2-minute match quiz to see what fits your background.
Take the 2-minute match quiz →
Retakes and the practice mode
| Rule | What Mercor states |
|---|---|
| Practice interview | Free, not tied to any application, does not affect future opportunities, never shared with companies, and you can take it as many times as you like |
| Official retakes | “Candidates are entitled to retake an interview up to three times across all applications related to that particular interview”, self-serve from the Assessments tab |
| Which attempt counts | Only the latest. “If you retake an AI interview or assessment, the previous submission is automatically replaced by the latest version” |
| A fourth attempt | By email only, with a screenshot or screen recording of the technical issue, and it is not promised |
| Reuse | “Each unique interview is only necessary to complete once; if another application necessitates the same interview, your existing result will be retained” |
That reuse rule cuts both ways and is worth reading twice. A weak interview follows you to every listing that uses the same one, and a strong one saves you from repeating it. It is an argument for doing the practice round properly.
What happens afterwards, including the silence
Mercor publishes a response time and then, for a large part of its catalogue, explains why you might never get one.
For listing-specific applications: “It usually takes between 2 to 4 weeks to hear back about whether you’ve moved on to the next stage.” There is no feedback on rejection, “due to the high volume of applications being processed”, and you cannot withdraw an application once submitted; it simply disappears from your dashboard “typically within a few weeks”.
For its standing talent-network listings, which are a large share of the 463 live roles on its worker platform, the vendor is blunt: “This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.” If you passed an interview and then heard nothing, that sentence is very likely your answer, and it is published rather than hidden. The same pattern shows up across this category, as we found writing Micro1 no response.
If a project does match, the stages are assessments or forms, sometimes a paid work trial, then an offer. Work trials are strict: one per person, paid once, and accepting one automatically rejects all the others you are holding. Onboarding then runs through Persona for identity, Certn or Zync for background checks, Insightful for time tracking, Okta for access, a contractor email on @mercor.expert, and for some projects a locked-down enterprise browser.
Support is itself largely automated: email goes first to an AI assistant Mercor calls Melvin, and in-project Slack support runs through one called Maven, which Mercor says “resolves around 80% of the requests it receives without escalating to a human”.
What does Mercor pay, and how?
Rates are published per listing and vary enormously by domain. Examples taken from live listings on September 17, 2026, quoted only as listed: generalist expert $40 to $70 an hour; bilingual Russian generalist $38 to $42; audiobook QA in German $15 to $20; electrical engineering $70 to $120; data scientist $100 to $150; medical expert $115 to $150; corporate lawyer in France $150 to $160; senior design expert $150 to $250. Listings phrase it as “most contracts in this field have paid” a given range, “set per project by scope and depth”. None of that is a forecast of what anyone earns.
| Mechanic | What Mercor states |
|---|---|
| Processor | Stripe Connect or Wise. “Mercor automatically assigns your payment provider based on your region. You do not choose it” |
| Schedule | Every Wednesday at about 12:00 PM PST, covering the previous work week, defined as Saturday 00:00 to Friday 23:59 IST |
| Currency | “All Mercor contracts are denominated in USD”, converted at payout |
| First payout | Stripe holds it for 7 days, a Stripe rule Mercor says it cannot override |
| Instant payout | Available for a 1.0% fee, arriving in about 30 minutes. The standard payout is free and takes up to 5 business days |
| Wise | You must action an email for each payout. “If you do not accept the email, the transfer stalls on Wise’s side. Mercor cannot push it through” |
| Minimum payout | Not published anywhere. We searched the full help centre |
| Tax | W-9 and a 1099-NEC over $600 in the US, W-8BEN outside it. You cannot contract through a company: “We are unable to accept W8-BEN-E” |
Mercor also states its own revenue model plainly, which is rarer than it should be: “Mercor applies a margin to companies’ pay rates once you accept a role… we only generate revenue when you are successfully placed.”
Three things worth knowing before you apply
1. There was a breach, and Mercor disclosed it properly. In a post dated June 25, 2026 the company said that “in late March 2026, Mercor was affected by a supply chain attack involving LiteLLM”, that “of our nearly five million experts, only a very limited subset had sensitive information affected”, that there was “no evidence that any of this data has been used fraudulently”, and that affected people were notified directly with an offer of identity protection. It named Mandiant as its forensics partner and listed its remediation steps. Seven civil suits were filed against Mercor.io Corporation in the Northern District of California in April 2026. We have not read those complaints, so we are not going to characterise what they allege, only that they exist and that the timing follows the breach window.
2. Its two country lists disagree. Mercor’s public supported-countries page names 156 countries and says it “does not engage contractors located in countries outside of those listed”. A separate list in its help centre carries 196 entries, including Venezuela and Yemen, which are absent from the first. Both are Mercor pages. If your country sits on one list and not the other, ask before you invest an evening.
3. A tension in its own documents, stated without accusation. Mercor’s website privacy notice says “we do not conduct solely automated processing of personal data for the purposes of evaluating… an individual’s personal aspects” and that consequently it does not offer rights over such automated decision-making. Its help centre says the AI interview is “the standard and only method for screening candidates” and that human feedback is “never utilized for making hiring decisions”. Those two positions are hard to hold at once. We are not alleging a breach of anything; we are noting that the privacy notice is scoped to the website rather than to the expert platform, and that the platform’s own privacy terms are not published publicly.
On the housekeeping side: Mercor’s help centre links to two of its own pages that return 404, including the one its “What’s Next?” card offers as interview preparation, and its help centre gives a corporate address that its SEC filings and job postings contradict. Minor, but worth knowing if a link you were told to follow does not resolve.
The honest read
For the platform itself rather than the interview, see our Mercor review and Mercor versus Micro1. For the wider field, Outlier alternatives is the broadest comparison we keep.
Screened, human-approved, filled ones deleted daily. No account, no paywall.
Frequently asked questions
What questions does the Mercor AI interview ask?
Mercor publishes none, and asks candidates not to share specific questions or screenshots. The questions are generated per role, and the same interview can be weighted differently by different listings, so a memorised question list would not help even if one existed. Any public Mercor question bank is either invented or leaked.
How long is the Mercor interview?
Mercor says most take approximately 20 minutes, though some are shorter or longer, and at least one listing specifies 25 minutes. An interview that does not meet the minimum completion criteria, such as duration, is not classified as completed.
Is the Mercor interview recorded?
Yes. Mercor retains the video and audio recordings and an AI-generated transcript, and uses them to evaluate your candidacy and recommend your profile. It does not provide candidates with copies of either the recording or the transcript.
Can I retake the Mercor interview?
Yes, up to three times across all applications that use that particular interview, and only your most recent attempt counts. A further attempt has to be requested by email with a screenshot or screen recording of a technical issue, and is not promised.
Is there a Mercor practice interview?
Yes, and it is free. It is not tied to any application, it does not affect future opportunities, it is never shared with companies, and you can take it as many times as you like.
Can I do the Mercor interview on my phone?
No. Mercor states that mobile browsers are not supported, and nor is Firefox. It recommends Chrome, and supports Edge and Safari. You need a working camera and microphone, and the interview will not start until the system checks pass.
How long does Mercor take to respond?
Two to four weeks for listing-specific applications, with no feedback given on rejection. Many of its listings are standing talent-network pools, and for those the vendor states outright that no decision is issued at all and that it reaches out only when a project matches, which can take from a week to several months.
How does Mercor pay?
Weekly, every Wednesday at about noon Pacific, covering the previous work week. Payment goes through Stripe Connect or Wise, assigned automatically by region rather than chosen by you, and all contracts are denominated in US dollars. Stripe holds the first payout for seven days, and Mercor publishes no minimum payout amount.
Mufy Pachorawala · Founder, getAIwork
My AI scans thousands of AI-job posts (42,827 screened so far) and I personally approve every listing before it reaches the board. I write these articles by the same rules: pay quoted only as listed, and no income promises. Read our editorial rules.
Related: Mercor review · Mercor vs Micro1 · Is Micro1 legit? · Micro1 no response · How to pass the Zara AI interview · The DataAnnotation Starter Assessment · Outlier alternatives
