Examopotamia1. Use a question bank, not a fixed paper
When every person gets a different selection of questions from a larger bank, a shared answer sheet stops being useful. Aim for a bank at least two to three times larger than the test.
2. Shuffle questions and answer options
Shuffling the order of questions and of each question’s options means “the answer to number 4 is C” no longer helps anyone sitting nearby.
3. Enforce the time limit on the server
A countdown in the browser can be paused or reset. A limit enforced by the server cannot: when time is up, late answers are simply not counted.
4. Keep answer keys out of the browser
Some quiz tools send the correct answers to the page so it can mark instantly. Anyone who knows how to open the page source can read them. Choose a tool that marks answers on the server.
5. Rotate question variants
Write two or three versions of important questions: same skill, different numbers or scenario. Candidates who share questions then share the wrong ones.
6. Limit attempts
One attempt per person (or per job application) removes “practice runs” with the real questions.
7. Watch integrity signals, don’t over-trust them
Switching tabs, pasting text or leaving the page are worth recording. Treat them as a reason to look closer or ask a follow-up question, not as proof of cheating: people get phone calls and lose connection.
8. Ask questions that need thinking, not searching
“What does VLOOKUP stand for?” can be searched in seconds. “Which formula would you use for this table, and why?” can’t. Scenario questions are harder to cheat on and better at measuring skill.
9. Consider open-book by design
If people will have reference material at work, let them use it in the test too, and set questions about applying knowledge under time pressure. Open-book tests remove the incentive to cheat.
10. Follow up in person
For hiring, a short conversation about two or three of the test questions quickly shows whether someone understood their own answers.
Examopotamia does 1–7 for you: question banks, shuffling, server-side timing and marking, AI variants, attempt limits and integrity scores. See how it works.