Guide

How online exam proctoring actually works

Proctoring software is sold as if it prevents cheating. It does not. What it does is produce signals, and the useful question is which signals you get, how reliable they are, and what decision you are entitled to make from them.

Three families, often sold as one

Browser-level integrity monitoring
The exam runs in a normal browser and reports events: the tab lost focus, the window blurred, fullscreen was exited, text was copied or pasted, developer tools opened. Cheap, private, and no install — but it only sees the exam window.
Recorded proctoring
Webcam, microphone and screen are captured for later review, sometimes with automated flagging. It sees the room, which the browser cannot. It also collects biometric data, needs consent and a retention policy, and demands bandwidth and a working camera from every candidate.
Live human proctoring
A person watches, in real time, and can intervene. The most reliable and the least scalable.

What a browser signal actually tells you

A focus-loss event says the exam surface stopped being the active window. It does not say why. A notification stealing focus, an accessibility tool, a flaky trackpad, and a deliberate search in another tab all look identical in the log.

This is why the count matters more than any single event, and why patterns matter more than counts. Four focus losses in a two-hour paper is noise. Twenty-two, each two to six seconds long, each immediately before an answer changes, is a pattern worth a conversation.

The failure mode to avoid is treating the log as a verdict. Automated action is reasonable for gross cases — a threshold that force-submits after a large number of violations — but a borderline count should route to a person, not to a rejection email.

What none of it can see

A second device is invisible to browser-level monitoring, and largely invisible to a webcam pointed at a face. So is a person off-camera. Any claim to prevent cheating outright is a claim about a room, and software running in a browser tab is not in the room.

The honest framing is deterrence plus evidence. Candidates who know signals are collected behave differently, and when something does go wrong you have a record to reason about.

Choosing by what is at stake

Practice and formative tests
Often no proctoring at all. The student is the beneficiary; surveillance adds friction and anxiety for no gain.
Mock tests in a series
Browser-level signals are proportionate. The score informs preparation rather than deciding an outcome.
Hiring screens
Browser-level signals, plus a later round where a person is present. Do not let an automated flag end a candidacy on its own.
Certification and final exams
A supervised session, an invigilated room, or recorded proctoring with a documented review process and a stated retention policy.

Telling candidates first

Say what is collected, what the threshold does, and what happens if it is crossed — before the attempt starts, not in a terms page. Candidates who understand the rules trip fewer signals accidentally, and a documented disclosure is what makes a later decision defensible.

Common questions

Can proctoring software stop someone cheating?

No. It deters, and it produces evidence. A second device or another person in the room is beyond what any browser-based or camera-based check can see, so treat proctoring as an input to a human judgement rather than as prevention.

Is browser-level monitoring enough for a high-stakes exam?

On its own, usually not. It is proportionate for practice and mock tests and for a first hiring screen. For certification or a final exam, add a supervised live session, an invigilated room, or recorded proctoring with a documented review process.

What integrity events does this platform collect?

Tab switch, window blur, fullscreen exit, copy, paste, right-click, and developer-tools open, plus a running violation count. Past a threshold you configure, the runtime can warn, force-submit, or ban. There is no webcam or screen recording.

More answers on the full FAQ.