Proctoring for Training Providers: When You Need It and When You Don’t
Most training providers need some exam security, but fewer need full proctoring than vendors suggest. The right level depends on what the result is used for, where the exam is taken, and what a compromised score would cost you. For low-stakes quizzes, good test design is often enough.
This guide gives you a simple risk-based way to decide on proctoring for training providers who run instructor-led (ILT) or blended courses. You will get a decision table, a worked cost example with labeled assumptions, and a checklist of low-cost controls you can use this week.
It is written for small and mid-size providers, not national testing programs. Where privacy or accessibility law is involved, it points you to the regulator and does not give legal advice.
Key takeaways
- Match security to stakes. A weekly quiz and a certificate that employers rely on need different controls.
- In-room delivery, randomized items and time windows stop a lot of cheating at almost no cost.
- Remote proctoring adds cost, privacy duties and accessibility duties, so use it only when remote delivery is truly needed.
- Research on remote proctoring is mixed and mostly from low-stakes settings, so do not treat any tool as proof.
- Write down your decision for each exam so you can explain it to learners, clients and auditors.
What does proctoring actually do, and what does it not do?
Proctoring is the supervision of an exam to confirm who is taking it and to deter or detect rule breaking. It can be a person in the room, a person watching a webcam, a recording reviewed later, or software that flags odd behavior. It supports security. It does not replace it.
Proctoring covers a few specific threats: impersonation, outside help, notes and unauthorized devices. It does nothing about leaked questions, a weak item bank or a badly set pass mark. That is why the ITC and ATP Guidelines for Technology-Based Assessment (November 2022) treat test security as its own chapter, “Technology-enabled test security”, and list remote online proctoring as one practice among many.
Think of proctoring as one layer. Item design, delivery rules and retake policy are the other layers, and for most ILT providers they do more work per dollar.
How do you choose proctoring for training providers? Four questions
Score each exam on four questions: the stakes of the decision, the delivery setting, the cost of a compromised result, and the learner population. Low answers on all four mean design controls are enough. High answers on stakes and cost mean you should add supervision. This is a practical rule of thumb, not a published standard.
- Stakes of the decision: Does the result only guide learning, or does it grant a credential, a license pathway or a job?
- Delivery setting: Is everyone already in one room with an instructor, or are people testing at home?
- Cost of a compromised result: What happens if someone passes who should not? Think safety, client trust, refunds and your own reputation.
- Learner population: How many learners, how spread out, and do any need accommodations, shared devices or low bandwidth?
Give each question a score of 1 (low), 2 (medium) or 3 (high). A total of 4 to 6 points to design controls only. A total of 7 to 9 points to supervised delivery. A total of 10 to 12 points to the strongest option you can run fairly. Treat the totals as a starting point and override them with judgment when one factor is extreme.
Which proctoring level fits which situation?
Match your score to one of five levels, from design controls alone up to automated flagging. Scores of 4 to 6 fit level 1, and 7 to 9 fit level 2. Scores of 10 to 12 point to levels 3 to 5 only when learners cannot test in person. Most ILT providers land on the first two.
| Level | Best fit | What it stops | Main drawback |
|---|---|---|---|
| 1. No proctoring, design controls only | Quizzes, practice tests, end-of-module checks | Answer sharing, memorized keys | Cannot confirm who took it |
| 2. Live in-room invigilation | Classroom exams, certificates tied to attendance | Impersonation, notes, phones, copying | Needs a trained person and a set time |
| 3. Remote live proctoring | High-stakes exams for learners who cannot travel | Impersonation and visible rule breaking | Highest per-exam cost, privacy and bandwidth issues |
| 4. Record and review | Medium to high stakes where live timing is hard | Visible rule breaking, with proof after the fact | Someone must review recordings; stored video is sensitive data |
| 5. Automated flags | A first filter on large remote volumes | Some unusual behavior, at scale | False flags need a human decision |
Level 5 should not make final decisions alone. Our earlier post on AI-driven cheat detection in online assessments covers how those tools work, so this guide does not repeat it.
Pro tip: Decide Before You Buy. Write one sentence per exam stating its score and level, before you look at any vendor. A written decision keeps a sales demo from raising your level, and it gives you something to show a client who asks why you did or did not proctor.
What does proctoring for training providers cost in time and money?
In-room invigilation costs staff time, and remote options add vendor fees, review time and support. Exact vendor prices vary, so this section uses labeled assumptions you should replace with your own numbers. The goal is to compare effort per candidate, not to quote a price.
Here is a worked example for in-room delivery. Assumptions: 40 candidates, tested in 4 sessions of 10, each session 2 hours, one invigilator paid $30 per hour, and 30 extra minutes per session for setup and ID checks. Staff time is 4 sessions x 2.5 hours = 10 hours. That is $300, or $7.50 per candidate.
Now the same 40 candidates under record and review. Assumptions: a vendor fee per exam (you must supply this, it is not estimated here), 25% of recordings flagged for human review, and 10 minutes of review per flagged recording at $30 per hour. Review time is 10 recordings x 10 minutes = about 1.7 hours, or $50, before the vendor fee and before any learner support time.
| Level | Staff time (40 candidates, example) | Extra costs you must add |
|---|---|---|
| In-room invigilation | 10 hours, about $300 at the assumed rate | Room, ID check process |
| Record and review | About 1.7 hours of review, about $50 at the assumed rate | Vendor fee per exam, learner tech support, video storage and deletion |
| Remote live | Proctor time is usually the vendor’s, billed in the fee | Vendor fee per exam, scheduling, learner tech support |
The point is the shape, not the totals. In-room cost is mostly your time and is predictable. Remote cost hides in fees and support. If you already run classes in person, in-room delivery is often the cheaper and simpler choice.
Which low-cost integrity controls work for in-person training?
The cheapest controls change how the exam is built and delivered, not who watches it. Randomized items, tight time windows and in-class delivery cut most casual cheating without any proctoring software. Use this checklist before you spend on supervision.
- Draw items at random from a bank. Build the exam from a larger item bank so neighbors do not see the same questions in the same order.
- Shuffle answer options where the wording allows it.
- Open the exam in a time window. Make it available only during the session, not all week.
- Limit item exposure. Retire or rotate items that have been used many times, and track which items a learner has already seen.
- Deliver in class. Use a room you control, with phones away and ID checked at the door.
- Set one attempt per window and a clear retake rule with a waiting period.
- Use scenario or application questions that are harder to look up than recall questions.
- Review results for odd patterns, such as identical wrong answers across seats.
Randomization only works if your bank is big enough. A bank barely larger than the exam gives every learner nearly the same test. Our guide to what an online assessment platform is explains item banks and delivery controls, and the SimpliAssess features page lists item banking and schedule management among its tools. It does not list proctoring, so check any platform’s published features against this list.
Pro tip: Build the Bank Before the Rules. Aim for at least two to three times as many items per topic as any one exam uses, then set exposure limits. The size and rotation rule are our suggestion, not a published standard, so test them against your own cohort sizes.
What does research say about remote proctoring?
The evidence is mixed and thin for high-stakes use. A 2025 randomized experiment found no significant overall effect of webcam proctoring on test scores, and it saw little visible cheating. But it was low stakes, so it cannot tell you what happens when a license or job is on the line.
The study, published in the Journal of Intelligence, had 226 participants recruited online and paid a fixed fee. Three participants (4%) showed clear cheating, such as taking notes or photographing the screen. The authors note that low stakes may limit how far the results apply to high-stakes testing.
Read that two ways. Where stakes are low, heavy surveillance may add little. Where stakes are high, we do not have strong public evidence either way from this study, so rely on layered controls and your own risk score. Be wary of vendor pages that quote large success figures without a named, checkable source.
What privacy and accessibility rules should you check first?
Before you use remote proctoring, check three things: biometric and video privacy rules where your learners live, your duty to give accessible exams, and how long you keep recordings. This is a starting list, not legal advice. Have a qualified lawyer review your setup.
On privacy, regulators treat face-based identification seriously. The UK regulator, the Information Commissioner’s Office, says biometric data is special category data when used for the purpose of uniquely identifying someone. If you test learners in the UK or EU, ask a specialist how data protection law applies to any face matching.
In the US, Illinois has a specific law. Its Biometric Information Privacy Act (740 ILCS 14) lists a “scan of hand or face geometry” as a biometric identifier, requires a written release from the person, and sets damages of $1,000 for negligent and $5,000 for intentional or reckless violations. Whether a given proctoring tool triggers it is a question for counsel.
On accessibility, the US federal rule for covered exam providers, 28 CFR 36.309, requires that an exam be administered so results reflect the person’s ability “rather than reflecting the individual’s impaired sensory, manual, or speaking skills,” in accessible facilities or with alternative arrangements. A webcam rule that blocks a learner with a disability is a risk. Plan an alternative path, such as in-person delivery, before exam day.
What should you do next?
Pick your three most important exams and score each on the four questions. Write the level next to each score, then apply the low-cost checklist before you add any supervision. Most providers find one or two exams that deserve more, and the rest need only better design.
If a credential is a revenue line, read how training providers sell certification exams, because the security plan is part of what makes an exam defensible. Then write a short learner notice that says what is recorded, how long it is kept and what the alternative is.
If you also need scheduling, delivery and tracking for classroom exams in one place, SimpliTrain’s team can show you how it handles them in a demo.
Frequently asked questions
Do I need proctoring for every training exam?
No. Weekly quizzes and practice tests rarely justify it. Use a risk score based on stakes, delivery setting, cost of a bad result and learner population. Reserve supervision for exams that grant a credential or support a safety, licensing or hiring decision, and use design controls for the rest.
Is in-person invigilation better than remote proctoring?
For learners already in your classroom, usually yes. It is simpler, cheaper to run and avoids webcam privacy and bandwidth problems. Remote proctoring makes sense only when learners cannot travel. Even then, offer an accessible alternative and keep recordings for the shortest time you can justify.
What is the difference between live, record-and-review and automated proctoring?
Live proctoring has a person watching during the exam. Record and review captures video and checks it later, often only flagged parts. Automated proctoring uses software to flag unusual behavior. Humans should make any final decision about a learner, because automated flags can be wrong.
Can I reduce cheating without proctoring software?
Yes. Draw randomized items from a large bank, shuffle options, open the exam only in a time window, limit item exposure and deliver in class with ID checks. Add scenario-based questions and review answer patterns afterward. These steps cost little and address the most common cheating.
Is remote proctoring legal for training providers?
Often yes, but it depends on where your learners live and what the tool collects. Biometric and video rules differ by place, and accessibility duties may apply. Check the regulator pages for each region you serve and get advice from a qualified lawyer before you start.
