Online Proctoring Software: How It Works, and What It Can't Do

Online proctoring software supervises a test taken away from an exam hall. Depending on the level, it locks down the browser, takes webcam snapshots, uses AI to flag anything unusual, or puts a live person on the other end of the camera. It exists for one reason: so that a result from a test taken at someone's kitchen table means as much as one taken under supervision.

We built proctoring into Zingalify, so we've had to make the trade-offs this guide describes, and we'll say where we landed.

The four levels of proctoring

No proctoring. Right for practice, lesson checks and anything where cheating only cheats the candidate.

Lockdown. The test notices when the candidate leaves the window or opens another tab, and can block copying. It's the lightest control that still deters casual cheating, and it costs the honest candidate almost nothing.

Automated (AI) proctoring. The webcam takes snapshots during the attempt, and software flags frames that look wrong: a second face, no face, a phone in view. Nothing is decided automatically; a person reviews the flags.

Live proctoring. A trained person watches in real time and can intervene. It's the strongest option and the most expensive, and it needs scheduling, which candidates in other time zones notice.

Snapshots or continuous video?

Some tools record the whole attempt as video. We chose webcam snapshots instead, included in every Zingalify attempt that uses AI proctoring. For most hiring and training tests, a series of snapshots is enough to review a flagged attempt, and it holds far less personal data than an hour of video of someone's home. For the highest-stakes exams, live human proctoring is available on our Enterprise plan.

AI flags, a person decides

The most important rule of automated proctoring: the software raises questions, a person answers them. A flagged frame might be a second person in the room, or a flatmate walking past, or a candidate with a disability who needs to look away from the screen. Treat every flag as a prompt to look, never as a verdict, and have a clear process for candidates to explain.

Setting it up so honest candidates don't mind

• Explain it before the test, in plain words: what the camera captures, who looks at it and when it's deleted.
• Offer a practice run with the same settings, so nobody discovers their webcam doesn't work at minute one.
• Plan for accommodations: extra time, breaks, screen readers, and candidates who need to look away from the screen.
• Give a way out for people without a webcam or a private space, such as a supervised slot or a different format.
• Mind time zones if you use live proctoring; a 3 a.m. slot isn't a real option.

What proctoring can't do

No proctoring catches everything. A determined cheat with a second device out of camera view, or a friend feeding answers through an earpiece, can beat snapshot proctoring and sometimes live proctoring too. That's why the strongest protection is the test design itself: large question pools so no two candidates see the same paper, randomised order, a sensible time limit, and questions that test applied judgement rather than facts you can search for. Our guide to creating an online test covers those.

Privacy and fairness rules

• Tell candidates first: what's monitored, how flags are reviewed, how long images are kept and who sees them.
• UK and EU: webcam images are personal data under GDPR; you need a lawful basis, a clear notice and a retention limit, and if you use facial recognition to identify people it becomes biometric data with stricter rules.
• United States: Illinois' BIPA and similar state laws regulate biometric identifiers such as face geometry, and the ADA requires accommodations, which may mean adjusting how flags are interpreted.
• Canada, Australia and India: PIPEDA, the Privacy Act and the DPDP Act respectively require notice, purpose limits and sensible retention for candidates' images.
• Everywhere: offer an alternative for candidates without a webcam or a private space, and never make a decision from an automated flag alone.

Hiring tests and exams need different settings

In an exam, the candidate wants the result and the institution needs it to be credible, so stricter supervision is expected. In hiring, the candidate is also judging you. Heavy proctoring on an early screen tells applicants you expect them to cheat, and the strongest ones have other options. For hiring, lean on test design and keep proctoring for the final, deciding test; for certification exams, proctoring is part of what makes the certificate worth having.

When you really need proctoring

Use it when the result alone decides something significant: a hire, a licence, a certificate. For an early hiring screen, question pools, random order, a time limit and a lockdown are usually enough, and they put fewer good candidates off. See our guide to choosing an online assessment platform for how proctoring fits the rest of the decision.

In Zingalify, proctoring is set per test, from none to live, and AI proctoring is included in the price of each completed attempt ($1 to $3 depending on the proctoring used), so there's no separate proctoring subscription.