Skip to content
Online proctoring

Know what happened during the attempt

26 integrity signals recorded per attempt, weighted into a single score, with the snapshots and timeline kept so every flag can be checked rather than believed.

  • Candidate consent required
  • Evidence retained
  • Human review queue

What it does

Observation, then judgement

The product's job is to give you a reliable record and a useful ranking. The verdict stays with you.

Webcam and screen capture

Periodic snapshots from both, retained against the attempt. A flag is always accompanied by what was actually seen.

Pre-flight verification

Before the timer starts: browser support, camera, microphone, screen share, connection speed and whether more than one display is connected — with plain instructions for anything that fails.

Identity verification

Optionally require a photo ID and a live capture, compared and recorded against the attempt, so the person who scored is the person you invited.

Weighted risk score

A 0–100 score with a low, medium or high band. Weights differ by signal, because a single blur is not the same as developer tools opening.

Attempt timeline

Every event in sequence with a timestamp, so you can see whether three flags were one moment of confusion or a sustained pattern.

Review queue

High-risk and flagged attempts collect in a queue for a human decision, recorded as cleared, confirmed or inconclusive with the reviewer's name attached.

Signals

Everything recorded during an attempt

Published in full. If a signal is not on this list, we do not collect it.

Identity and presence

  • No face visible in frame
  • More than one face detected
  • Camera switched off mid-attempt
  • Gaze directed away from the screen
  • Sustained head movement away from the screen

Environment

  • Phone or handheld device in frame
  • Speech detected while alone
  • Background noise above threshold
  • Earphones or headphones detected
  • Additional display connected

Browser and device

  • Tab switched or window lost focus
  • Window minimised
  • Full screen exited
  • Developer tools opened
  • Screen share stopped
  • Page refresh attempted
  • Screenshot attempted
  • Browser extension detected
  • Automation or remote-control tooling detected

Answer integrity

  • Copy, cut or paste inside the assessment
  • Navigation to a restricted site
  • Blocked action attempted during lockdown

Live monitoring

Intervene while it still matters

An invigilator console shows every active attempt with its progress and flags as they occur, so a problem is handled during the assessment rather than discovered after it.

  • Message a candidate directly without leaving the console
  • Pause an attempt, or extend time when a genuine technical fault costs someone minutes
  • Force-submit a section, or the whole assessment, when a session has to close
  • Watch flags accumulate live rather than reading them the next morning

Why this matters

Most integrity problems in a large drive are not misconduct — they are a candidate whose camera dropped, or whose network died at minute forty. Seeing that live lets you fix it. Seeing it in a report the next day means you have already scored someone unfairly.

Fairness

What proctoring cannot do

Worth being direct about, because the alternative is a customer who trusts the score more than it deserves.

Does proctoring decide whether a candidate cheated?
No, and we would not trust a product that claimed it did. It records what happened and weights it into a score to tell you which attempts are worth a look. The decision is a person's, on the evidence, and the review outcome is recorded with their name on it.
Can it stop a candidate from cheating?
It raises the cost and makes attempts visible: full-screen is enforced, copy and paste are recorded, developer tools and additional displays are detected, and randomised question pools mean candidates in the same room see different papers. Anyone claiming a browser can prevent cheating outright is overselling — a page cannot inspect the room or the second device.
What about false positives?
They happen — a family member walking past triggers a multi-face signal. That is exactly why the score links to its evidence and why flagged attempts go to a human queue rather than an automatic rejection. A camera failing to capture is recorded as a diagnostic, not as misconduct.
Do candidates know they are being monitored?
Yes, explicitly. The system check tells them what will be captured before anything is recorded, and they cannot start without granting camera and screen-share permission themselves. Covert monitoring would be both wrong and, in most jurisdictions, unlawful.
Can we run assessments without proctoring?
Yes. Proctoring is configured per assessment. A take-home exercise or an internal skills check can run with it off entirely.
Can candidates take an assessment on a phone?
Proctored assessments require a desktop browser — the monitoring signals and the code editor both depend on capabilities phones do not reliably provide. The system check tells a candidate this up front rather than failing them halfway through.

Try it on a real attempt

Run a proctored assessment on the free trial and look at the evidence it produces before you decide.

Have questions? Email sales@parikshafy.com