Facial Validation gives you a record of each unsupervised test session through the candidate's webcam. It helps you check that the right person took the test, on their own, and that they had a fair chance to show what they can do.
It compares snapshots taken during the test with a reference image the candidate provides at the start. It isn't facial recognition, it doesn't use external databases, and it makes no decisions: it flags what a person should review. This guide explains how to read a portfolio, when to follow up, and how to talk to candidates fairly.
What Facial Validation Records
Reference Image
Captured with consent at the start, and used only for that session.
Snapshots
Taken at intervals throughout the assessment.
Face Presence and Position
Whether a face is visible, and roughly where it sits in the frame.
Match Confidence
How closely each snapshot matches the reference image.
Session Portfolio
All of the above in one place, for you to review.
What It Can Help You Spot
- Help from another person
- Someone else taking the test
- A camera that is covered, moved or shown a still image
- Technical problems that affected the session
- Distractions or changes in the room that may have affected performance
Four Things to Look At
Most anomalies have an innocent explanation. Each indicator below shows what is normal, what usually explains a blip, and what deserves a closer look.
Face Visibility and Position
- What to Look For
- The face clearly visible for most of the session, broadly facing the screen, at a reasonable distance.
- Often Innocent
- A badly placed camera, technical problems, natural movement while thinking.
- Worth a Closer Look
- Long periods with no face; repeatedly looking away for long stretches; consistently extreme angles; big changes in distance from the camera.
Match Confidence Over Time
- What to Look For
- Whether match confidence stays steady, drifts gradually or drops sharply.
- Often Innocent
- Poor lighting, a low-quality camera, natural movement.
- Worth a Closer Look
- Consistently low confidence; a sharp drop towards the end of an ability test; very low confidence that suggests a different person; no change at all between images.
Consistency and Surroundings
- What to Look For
- Whether the candidate's appearance, position and background stay broadly the same.
- Often Innocent
- A frozen camera, a software glitch, changing light, household activity, moving to another room.
- Worth a Closer Look
- The same image repeated throughout; big changes in the room mid-test; drops in confidence that coincide with those changes.
Patterns Across the Test
- What to Look For
- How behaviour on camera lines up with progress through the test and the type of question.
- Often Innocent
- Looking away to think during harder questions is normal.
- Worth a Closer Look
- Absences or looking away that follow a systematic pattern, especially if it matches the hardest questions.
Three Levels of Concern
Minimal Concern
The face is consistently visible with strong matches, position varies naturally, surroundings are stable, and any minor anomalies are easily explained.
Warrants Review
Occasional drops in confidence or positioning problems, or some inconsistency that could be technical. Look at the context and consider talking to the candidate.
Significant Concern
Frequent or long absences, several red flags together, or patterns that point to outside help or a different person. Follow up promptly; a supervised retest is likely.
Was It a Fair Test?
Facial Validation is not only about misconduct. Even when nothing is wrong, ask whether the candidate had a fair chance to perform at their best.
Signs of Good Conditions
- Stable surroundings throughout
- Steady match confidence, suggesting an undisturbed session
- A settled, comfortable position
- Timing that suggests genuine engagement
Signs of Poor Conditions
- Changes in the room that suggest household disruption
- Confidence swings that suggest interruptions
- Repeated repositioning that suggests discomfort
- An unsuitable place, such as a busy office or public transport
Facial Validation helps you ask better questions. It never answers them for you.
Flags are indicators, not evidence of misconduct. A trained person reviews every flagged session in context, hears the candidate's explanation, and stays responsible for the outcome.
Talking to the Candidate
Before You Speak to Them
- Review the whole portfolio systematically.
- Note the timestamps and confidence levels you want to ask about.
- Consider technical explanations for what you see.
- Prepare specific, non-accusatory questions.
- Have a retest option ready in case you need it.
Opening the Conversation
During your recent assessment, our monitoring system picked up some irregularities that we'd like to discuss with you. This doesn't necessarily mean anything was wrong, but we do need to understand what happened during your session.
Questions You Might Ask
Technical Problems
- Did you have any technical difficulties during the assessment?
- Did you have any problems with your camera or lighting?
- Did you need to adjust your camera or position during the test?
Setting and Surroundings
- Can you describe where you took the test, and what device you used?
- Was anyone else in the room during the assessment?
- Did you stay in the same place throughout, or did anything change?
Specific Anomalies
- There were long periods where your face wasn't visible to the camera. Can you help us understand why?
- The system detected other people in the camera frame. Can you tell us about that?
- How clearly the camera could see you changed a lot during the test. What might have caused that?
- Do you feel you were able to show your best performance in this assessment?
Weighing the Explanation
Judge whether the explanation fits the evidence, not how it is delivered. Nervousness, hesitation, an unusual communication style or trouble recalling details are common among candidates who have done nothing wrong, including neurodivergent candidates and people speaking a second language.
Usually Credible
- Technical problems: a poor connection, a camera fault, software issues
- Surroundings: poor lighting, a shared space, household activity
- Physical needs: a drink, the toilet, a medical need
- Setup: repositioning the camera or moving somewhere quieter
- Everyday interruptions: family, pets, deliveries, phone calls
Needs Further Discussion
- The explanation doesn't account for what the portfolio shows
- It conflicts with the timestamps or other evidence
- Details of the same event change when revisited
- There is no explanation for a clear, sustained anomaly
What to Record
- The candidate's explanation, in their own words
- Your view of whether it fits the evidence, and why
- Any further evidence from the conversation
- Next steps: accept, retest or review further
- The date and time of each interaction
Recommended Actions
Accept the Result
Minor anomalies with explanations that fit the evidence, and no pattern of concern.
Accept the result and record what you found.
Accept With a Note
Reasonable explanations, but distractions or technical problems may have affected performance.
Accept the result with a note on possible impact. If the role is critical, consider offering a supervised retest.
Supervised Retest
Significant anomalies that remain unexplained, or explanations that don't fit the evidence.
Ask the candidate to retest under supervision, and record your concerns.
Formal Review
Several red flags together, with explanations that do not hold up against the evidence.
Follow your organisation's procedure for a formal review, and arrange a supervised retest.
Six Common Scenarios
| What You See | Likely Causes | Follow Up By | More Concerning If |
|---|---|---|---|
| Low match confidence throughout | A poor camera, bad lighting or technical problems | Discussing the setup; offering a supervised retest in better conditions | The data shows problems the candidate can't account for |
| Face missing during harder questions | Natural thinking behaviour, notes, or outside help | Asking about distractions and how they worked through the questions | Absences systematically follow question difficulty |
| Surroundings change mid-test | Household interruptions, changing light, moving rooms, or outside help | Asking what happened and why the setting changed | Changes line up with particular questions or a jump in performance |
| Position keeps changing | Discomfort, device adjustments, distractions | Discussing the setup and their comfort | Changes follow a pattern linked to the questions |
| Identical images throughout | A frozen camera or software, or a still photo | Checking which device and camera were used | The images are perfectly identical, with no sign of a fault |
| Large swings in match confidence | Lighting, camera adjustments, changes in the room | Asking about lighting and setup | Patterns suggest different people or deliberate interference |
Fairness and Good Practice
Human Oversight
A trained person reviews every flag in context, weighs the explanation and makes the decision. Reviewers should be trained to recognise false positives.
Transparency
Explain the monitoring to candidates before they start, set clear expectations, offer reasonable adjustments, and document every decision.
Avoiding Bias
Allow for cultural differences in eye contact and posture, disabilities that affect position or movement, varying confidence with technology, and the fact that not everyone has a quiet place to test.
Data Protection
Images are stored securely, access is limited to authorised staff, retention periods are clear, and candidates' privacy rights are respected throughout.
Limitations
Equipment and Conditions
Camera quality, lighting, device and connection all affect match confidence.
False Positives
Natural movement, technical glitches, surroundings outside the candidate's control and cultural norms can all produce flags.
Never Used Alone
Read Facial Validation alongside the Confidence Score, the candidate's background and your other recruitment evidence. Interpreting Confidence Scores
The Evidence Behind It
In matched field data, sessions with Facial Validation showed significantly higher behavioural integrity than sessions without it, making it the strongest single layer of Podium's integrity framework.
Read the Integrity Research