AI Proctoring

A Practical Guide to Live Exam Monitoring at Scale

05 May 2026 · 1 min read
A Practical Guide to Live Exam Monitoring at Scale

Monitoring a live exam for 50 candidates is straightforward. Monitoring 5,000 simultaneously is an operations challenge as much as a technology challenge. This guide covers both.

Technology layer

At scale, the invigilator-to-candidate ratio matters. Human live monitoring is typically 1:30 (one invigilator can watch 30 live feeds). AI monitoring changes this to effectively 1:unlimited, with human review triggered only when the AI flags an incident.

The triage dashboard

ExamRankers' live monitoring dashboard shows all active candidate feeds in a grid. Feeds with AI flags are automatically surfaced to the top of the grid and highlighted. Invigilators focus their attention where it matters most — not watching 5,000 peaceful candidates.

Escalation protocols

Define your escalation protocol before the exam, not during it. Common protocol: AI flags → invigilator review → if confirmed, exam paused and candidate messaged. For serious violations, exam terminated and incident logged for review committee.

Communication during the exam

ExamRankers includes a direct messaging system between invigilators and candidates. For ambiguous flags (e.g. a power cut that caused a tab switch), the invigilator can message the candidate, note the explanation, and dismiss the flag — all without interrupting the exam flow.

Post-exam review

Schedule a review window of 2–3 hours immediately after the exam ends. Review all High severity flags before releasing results. Low severity flags can be reviewed in bulk with a single dismiss-all action if the overall pattern doesn't suggest coordinated malpractice.

More Articles