Online Assessments

Beginner · 16 min read · Interview Playbook, Interview Playbook

Timed tests on HackerRank and CodeSignal: scoring, partial credit, input handling, hidden-test traps and time management.

Before most coding interviews there's an online assessment (OA): a timed test on a platform such as HackerRank, CodeSignal or Codility, taken alone, with an automatic score.

This lesson is about the parts of an OA that differ from a live interview: format, scoring, input handling, time management and the tests you can't see.

What an OA usually looks like

Some platforms, notably CodeSignal, produce a standardised score that several companies accept, so one good attempt can be reused for months. Others are one-off tests per company.

How you are scored

So the goal is maximum total tests passed, which changes the strategy compared with a live interview.

Strategy: the first five minutes

Strategy: each problem

The tests you can't see

When a solution passes the samples but fails hidden tests, it's almost always one of these:

Input and output: read the format yourself

Many OAs, especially on HackerRank, give you stdin and stdout, not a function signature. You parse the input yourself. Fumbling this costs more time than most algorithms. Practise this template until it's automatic:

Work simulations and work-style surveys

Some OAs, Amazon's in particular, include scenario and survey sections. They're scored, and people do fail them.

Before the day

Interview questions

  • You have 25 minutes left, one problem fully solved and two untouched. What do you do?
  • Your solution passes all sample tests but only 7 of 15 hidden tests. What do you check first?

Practice

Write your answer before opening the suggested one. Anything you mark "Partly" or "Missed it" comes back in your review deck.

Further reading

Key takeaways

  • OAs are scored by hidden tests, usually with partial credit: maximise total tests passed.
  • Read all problems first, order by confidence, and time-box each one.
  • Submit a brute force early when the optimal idea isn't clear, then improve it.
  • Most hidden-test failures are overflow, time limits, edge cases or output format.
  • Practise reading stdin with buffered input and printing with a flushed writer.
  • Work simulations and surveys are scored too: answer consistently with the company's values.
  • Never use outside help unless allowed; proctoring and plagiarism checks are standard.

Previous: Solving Out Loud · Next: Spot-the-Pattern Drills