In this article
Before you start
This page is one part of the interview process, stage by stage, where the online assessment is usually the first filter.
Why interviewers ask this
A candidate who misreads an online assessment can lose the process before any person meets them. One third-party guide (titled 2021, so check the current format) describes the result of the Amazon online assessment as the thing that decides whether a candidate moves on to phone or in-person interviews (AlgoMonster, accessed 2026-10-26). Recruiters and interviewers therefore want to hear that you know three things: the test has no live interviewer, the grade comes from cases you cannot see, and the employer may track how you work.
A weak answer says "it is a coding test online." A strong answer names the automatic grading, the hidden cases and the time limit. A Staff-level answer adds that formats differ by company and that the rules are the employer's choice, so you ask the recruiter what yours contains.
Online assessment from first principles
- A hidden test case is a test case the candidate cannot see while solving, which the platform runs against the submitted code to check edge inputs and large inputs that the visible samples do not cover.
- Plagiarism flag
- A plagiarism flag is a marker a testing platform places on a candidate's report when submitted code is highly similar to another candidate's code, meant as a signal for a person to review, not as proof.
In plain words
Think of a driving theory exam marked by a machine. You sit alone, the clock runs, and a sealed answer key decides your score. Nobody asks you to explain your thinking. The analogy breaks in one place: an online assessment can also record behavior, such as tab switches, and the employer chooses which records to switch on. Most people who search this term mean the coding version, but the same words also cover other kinds of test, listed in the next section.
In one precise sentence, an online assessment is an automatic, timed screen whose score comes from hidden test cases and whose rules the employer sets.
What types of online assessment are there?
Online assessments come in several common types, and the grading differs by type. A coding test runs your code against test cases and scores the share that pass. An aptitude or reasoning test asks numerical, verbal or logical questions and is usually scored against an answer key.
A work-style or situational judgment test asks how you would act at work and is scored against the employer's own profile; one guide describes an Amazon stage as a work-style assessment with no coding (AlgoMonster). A recorded video assessment asks you to record answers to set questions, which a person or a tool reviews later. Software roles lean on coding tests, while many other fields lean on the other three. Which types you get is the employer's choice, so read the invitation.
Step 1: The employer sets the rules and the timer
The employer, not the platform, decides the rules of an online assessment. The employer picks the problems, the timer and the integrity settings in a testing platform; HackerRank, CodeSignal and Codility are common ones. Proctoring is the monitoring of a test taker's behavior during a test, such as leaving the window. HackerRank's guide for employers says copy-paste tracking is on by default for customer tests. A vendor blog post says its tab proctoring detects when candidates switch away from the test window and that its secure mode enforces full-screen testing.
Time varies. One guide describes Google's online assessment as two to three questions in 90 minutes (IGotAnOffer), and your own limit is whatever your invitation states.
Input: an invitation link and a deadline. Output: a timed session with rules you can read first. You now know why the invitation matters: it is the only place your rules are written down.
Step 2: Hidden test cases decide the score
Hidden test cases decide the score of an online assessment. The platform runs your submitted code against cases you never saw, and the code is graded automatically, with no person reading it during the test. The visible samples only show that your code runs on easy inputs.
Hidden cases usually probe two things: unusual inputs, such as an empty list, and large inputs that make a slow method run out of time. Suppose a solution passes all four visible cases and 14 of 20 hidden cases (illustrative). It then holds 70% on hidden cases, and the employer's weighting decides what that means.1 Platform cut-offs are set by each employer and are not public, so no page can honestly state a passing mark.
Input: your code. Output: a score computed from every case. You now know why a solution that "works on the examples" is unfinished: write tests for empty, single-item and very large inputs before you submit.
Step 3: A report goes to the employer
After grading, the platform builds a report for the employer. HackerRank's guide is written for employers and describes tracking such as copy-paste activity, and it describes how code similarity is flagged: matches begin at 75% similarity between candidates' attempts, and at 90% when a candidate wrote fewer than 10 lines (HackerRank, accessed 2026-10-26). The flag marks similarity between candidates' code; it does not itself say why the code matches.
Input: your submission and the recorded behavior. Output: a report read by people at the employer. If your score is strong enough, the next stage is a live round; the wider process shows what usually follows.
Step-by-step description
- The employer sets the rules and you solve.
- Hidden test cases grade the code.
- A report goes to the employer.
- The employer sets the timer and rules before the test starts.
- Hidden test cases grade the submitted code automatically.
- A report with the score goes to the employer.
Worked example: reading 14 of 20 hidden cases
Take one problem with four visible cases and 20 hidden cases, all numbers illustrative.1 A solution passes all four visible cases and 14 of the 20 hidden ones. Of the six failures, assume two are edge-case errors and four are time-outs, because the code uses a nested loop over all items. At 100,000 items, a nested loop over all items runs 100,000 × 100,000 = 10,000,000,000 times, while one pass takes 100,000 steps.1 The table below puts the same submission on three scales, so you can see how much the choice of scale moves the result.
| Measure | Value (illustrative) |
|---|---|
| Visible cases passed | 4 of 4 (100%) |
| Hidden cases passed | 14 of 20 (70%) |
| Score if all 24 cases weigh the same | 18 of 24 (75%) |
| Hidden failures from edge cases | 2 |
| Hidden failures from time-outs | 4 |
| Steps in a nested loop over 100,000 items | 10,000,000,000 |
| Steps in one pass at 100,000 items | 100,000 |
If a test has two such problems and the second scores fully, the average is 85%, again illustrative.1 The fix for the four time-outs is a better method, not more typing.
Text description of the diagram
- A solution can pass all four visible cases and only 14 of 20 hidden cases.
- That is 70% on hidden cases and 75% if all 24 cases weigh the same.
Knowing it is half the job. Saying it is the other half.
Interview Clozo is an AI interview copilot for real interview rounds: it hears the question and puts a strong answer on your screen, and you say it in your own words. See the features or how it works.
Start Free Trial →The Staff-level answer
What is an OA interview?
An online assessment is an automated, timed test an employer sends early in hiring, usually coding, run on a platform such as HackerRank, CodeSignal or Codility. No interviewer watches live, and hidden test cases grade the code, so I write for edge cases and large inputs, not just the samples. I read the employer's rules first, because copy-paste and tab tracking can be switched on and a person reads the report. Formats differ by company, so I ask the recruiter how many problems, how much time, and whether tools are allowed.
What makes it Staff-level: it names the grading mechanism, the integrity settings and the unknowns you would confirm, instead of only defining the test.
The rubric
| Dimension | Weak answer | Strong answer | Staff answer |
|---|---|---|---|
| Definition | "An online coding test." | A timed, automated test sent early in hiring with no live interviewer. | Adds that it is an early filter whose result decides who reaches live rounds. |
| Grading | Says it checks the sample output. | Says hidden test cases grade the code automatically. | Explains edge and large inputs and why employer weighting changes the meaning of a score. |
| Rules | Assumes nobody can see what you do. | Knows employers can track copy-paste and tab exits. | Reads the invitation first and treats the rules as the employer's choice. |
| Variation | Says every company is the same. | Notes that format varies by company and role. | Asks the recruiter for problem count, time and tool policy before preparing. |
| Next step | Stops at the test. | Mentions the live rounds that follow. | Connects the score to the process and prepares the follow-up round. |
Three variants of this question
How is an OA different from a phone screen?
A phone screen is a live conversation with an interviewer, while an online assessment is a remote test with no verbal interview (IGotAnOffer). Only the phone screen lets you explain your thinking out loud.
What is the difference between an OA and a take-home?
An online assessment is short, timed and auto-graded on a platform, while a take-home is a longer project a person reads. See the take-home assignment example for the second format.
What if the OA allows AI tools?
The policy depends on the employer. Read the invitation, ask the recruiter, and see the AI-assisted coding interview for a format where tools are part of the round.
Common mistakes
- "I passed all the sample tests, so I passed." (wrong) The grade comes from hidden cases. (right) "I tested empty, single and very large inputs before submitting."
- "Nobody can see what I do on my screen." (wrong) Employers can switch on copy-paste tracking, tab proctoring and secure mode. (right) "I read the rules in the invitation and work inside them."
- "A similarity flag means automatic rejection." (wrong) The vendor describes it as a similarity flag, and what follows is the employer's choice. (right) "A flag is a signal for review; I write my own solution."
- "Every company runs the same OA." (wrong) Formats differ by company and role. (right) "I ask the recruiter for the format before I prepare."
Follow-up questions
What is oa in interview?
In an interview process, OA stands for online assessment: an online remote test with no verbal interview, given early in the process (IGotAnOffer). It is usually coding and sometimes aptitude or work-style questions, and one guide says performance in it decides whether the profile proceeds to interviews (AlgoMonster).
Can you fail an OA after passing the sample tests?
Yes, because the score comes from hidden test cases that check edge inputs and large inputs the samples do not cover. A solution that is correct but too slow can fail the time-limit cases.
Is a plagiarism flag the same as being rejected?
No. HackerRank describes the flag as a marker of code similarity between candidates, and what the employer does with it is the employer's decision (HackerRank).
How long is an online assessment?
It depends on the employer. One guide describes Google's online assessment as two to three questions in 90 minutes (IGotAnOffer); your invitation states your own limit.
Does every company use the same OA format?
No. AlgoMonster describes a multi-part Amazon test with a debugging stage, a LeetCode-style coding stage and sometimes a work-style assessment, while another source describes Google's as a virtual coding test (IGotAnOffer). For specific companies, see OpenAI interview questions and process, the Anthropic interview process and the Amazon applied scientist process.
- An online assessment is a timed, automated test sent early in hiring, with no live interviewer.
- Hidden test cases decide the score, so passing the visible samples is not enough.
- The employer sets the time limit and integrity settings, such as copy-paste and tab tracking.
- A similarity flag is an indicator for review, not proof of wrongdoing.
- Formats differ by company, so ask the recruiter what your test contains.
Sources
- Plagiarism Best Practices Guide — HackerRank Knowledge Base (accessed 2026-10-26; undated page), 26 Oct 2026
- Blocking ChatGPT in Real Time: Tab-Proctoring and Secure-Mode Workflows — HackerRank vendor blog (accessed 2026-10-26; undated page), 26 Oct 2026
- Amazon Online Assessment Questions Preparation — AlgoMonster (secondary, third-party; accessed 2026-10-26; undated page), 26 Oct 2026
- Google Online Assessment Guide — IGotAnOffer (secondary; accessed 2026-10-26; undated page), 26 Oct 2026
