Reviewing Candidate Proctoring Reports

Updated 1 month ago

Maintaining the integrity of technical assessments is crucial for making fair hiring decisions. ForceHQ provides comprehensive proctoring tools. This guide explains how to interpret the data in a candidate's proctoring report when flags are raised.

Understanding the Proctoring Trust Score

At the top of every assessment report, you will see a Proctoring Trust Score out of 100.

  • 90-100: High confidence. No significant anomalies detected.
  • 70-89: Medium confidence. Minor anomalies detected (e.g., brief loss of focus), but likely not severe cheating. Review is recommended.
  • Below 70: Low confidence. Significant suspicious activity detected. Manual review is absolutely required before advancing the candidate.

Reviewing Specific Flags

If the trust score is low, scroll down to the Proctoring Timeline. This is a chronological log of all flagged events during the assessment.

1. Browser Tab Switches

If Browser Lockdown is enabled but the candidate managed to bypass it, or if it's set to "Soft Warning" mode, you will see flags for "Focus Lost". The system logs exactly how many times the candidate navigated away from the assessment tab and for how long. Brief switches (1-2 seconds) might be accidental, but frequent, long switches strongly indicate searching for answers externally.

2. Face Detection Anomalies

If Webcam Monitoring is enabled, the AI constantly analyzes the video feed.

  • Multiple Faces Detected: This flag triggers if a second person enters the frame. The report will include the specific photo snapshot where this occurred.
  • Face Not Found: Triggered if the candidate leaves the computer entirely during the test.
  • Face Mismatch: Triggered if the person taking the test suddenly changes halfway through.

3. Copy-Paste Violations

In coding assessments, pasting large blocks of code from the clipboard is often a red flag. The system logs every paste event, including the exact text that was pasted. Review these logs carefully; pasting a standard boilerplate function is normal, but pasting the exact solution to the proprietary algorithmic challenge is a clear violation.

Making a Decision

Technology provides the data, but humans must make the final call. When reviewing flagged reports, consider context. Was a "background noise" flag just a dog barking, or someone whispering answers? Use the provided audio snippets and screenshots to investigate.

If you determine a candidate has violated the integrity policy, you can use the Invalidate Result button within the report to mark the assessment as failed and notify the candidate.

Was this article helpful?

Ready to Fix Hiring at Scale?

Build a structured hiring system that scales without slowing teams down. Join the companies already hiring smarter, faster, and better with ForceHQ.

Book a demo