Skip to main content
All CollectionsGuide for RecruitersResources
Recruiter Reference: Virtual Proctoring
Recruiter Reference: Virtual Proctoring

Use HiringBranch Virtual Proctoring to ensure test integrity

Updated over a year ago

HiringBranch Virtual Proctoring uses AI-powered face detection to monitor candidates during their assessment. Potential integrity issues are automatically flagged for review.

Contact us if you're interested and we can get you started.

Options

Proctoring can be enabled on selected assessments on your site. In a proctored assessment, you can choose to enable any or all of the following:

  • before starting the assessment, require candidates to provide a reference photo of their face using their webcam

  • during the assessment, monitor the candidate to ensure that one and only one face is visible to the webcam

  • monitor the candidate to ensure that they do not change their browser's window or tab

Requirements

To complete a proctored assessment, candidates will need the following system setup:

  • a desktop computer, an Android phone or an iOS device

  • a functioning webcam

  • a browser with permission to access the webcam.

Completing a Proctored Assessment

If a candidate has trouble with proctoring at any point during the assessment, they can contact our support team through the help icon on the lower left corner of the screen.

1. Disable any virtual cameras

If a candidate has a virtual camera (Snapchat, OBS, etc), thy must disable it before they can start the assessment.

2. Allow the webcam

When a candidate first opens a proctored assessment, they will be asked to give their browser permission to access the webcam. If the candidate does not give access to the webcam, they will not be able to start the assessment.

3. Take a reference photo

If required, the candidate will then be asked to provide a reference photo. After candidate centers their face in front of their webcam, they will be able to click OK to take the photo and begin the assessment.

4. Keep one face visible

If required, the candidate must keep their face visible to the webcam during the entire assessment. This means staying close to the position they had when they took the reference photo. Shifting around, chin-scratching and tea-sipping are allowed. If a face is not detected, the system will raise a flag on the candidate's report.

The candidate must not allow another face to become clearly visible to the webcam. Photos in the background and people walking past in the distance will not produce flags. However, if another person approaches the screen to a readable distance, the system will raise a flag.

5. Do not change windows or tabs

If required, the candidate must not change windows or browser tabs during the assessment. Any window change will produce a flag on the proctoring report.

Reporting

You can access the proctoring report in two ways:

  • on the Reports page, choose the Proctoring option. Candidates who have too many flags on their assessment will receive a red flag icon. Click the icon to open the proctoring report.

  • from the user edit page, open the Courses tab and click Proctoring.

The proctoring report shows course and candidate information, including:

  • the reference image taken at the start of the assessment (if required)

  • a message showing if the candidate has too many flags or not

  • a table of all flags generated for that candidate's assessment.

If you believe that a flag was falsely raised, you can delete that flag from the report. Also, if you believe that a flag was not raised when it should have been, you can manually create a flag.

The platform cannot automatically determine if a candidate's assessment should be discarded based on the proctoring results. It can only determine if the number of flags generated warrant further investigation by the assessment reviewers.

About Face Detection

The face detection algorithm can only detect the number and position of faces that are in view of the candidate's camera. The model's accuracy has been verified using a wide variety of faces from around the world. Any form of identity recognition is explicitly beyond the model's capabilities. The model cannot detect any personally identifiable information, including but not limited to gender, ethnicity, or age.

Flags raised by the model cannot automatically discard a candidate's assessment. The final decision always and exclusively remains in the hands of assessment reviewers.

Did this answer your question?