Last verified: 2026-09-19
TL;DR
Online proctoring (live remote monitoring, AI-driven automated review, or a blend of both) and in-person testing at a physical center each protect exam integrity through different mechanisms: continuous human or algorithmic observation versus a controlled physical room. The right choice depends on the stakes of the credential, the geography and technology access of the candidate pool, and the regulatory or accreditation requirements attached to the certification. Most credentialing bodies now run a hybrid model, offering both delivery methods for the same exam rather than picking one exclusively.
What Do Online Proctoring and In-Person Proctoring Mean for Certification Exams?
Proctoring is the supervision mechanism that a credentialing body uses to verify that the person taking an exam is who they claim to be and that they are completing it without unauthorized assistance. Every certification exam needs some form of it, because the value of a credential rests entirely on the assumption that the person holding it actually passed the exam under fair conditions. Without a defensible proctoring method, a certification's market value erodes quickly, since employers and licensing boards stop trusting the credential.
Online proctoring refers to remote exam supervision conducted through a webcam, screen-share, and browser lockdown software, using either a live human proctor watching in real time, an AI system flagging suspicious behavior for later human review, or both layered together. In-person proctoring means the candidate travels to a physical location, often a dedicated test center or a supervised room at an employer or educational site, where a human proctor watches the room directly. Both methods aim to solve the same problem: preventing impersonation, unauthorized materials, and outside assistance. They differ in where the supervision happens and what infrastructure it depends on.
This decision matters to credentialing bodies and associations because it touches accreditation compliance, candidate satisfaction, program cost structure, and the defensibility of the credential if a score is ever challenged. A poor proctoring choice can suppress candidate volume (if the process is too burdensome), invite legitimate security challenges (if it's too loose), or trigger accreditation findings from bodies that require documented exam security controls, such as those aligned with ANSI/ISO/IEC 17024 for personnel certification programs.
Four Proctoring Models Certification Programs Use
Four distinct models cover most of how certification programs proctor exams today, and each one optimizes for a different variable at the expense of another.
Live remote human proctoring puts a trained proctor on a video call with the candidate for the full exam session, watching in real time and intervening if something looks wrong. This model optimizes for immediacy and human judgment; a live proctor can catch context that automated systems miss, like a second person entering the room. The tradeoff is staffing cost and scheduling: a proctor is typically tied up for the full length of one exam session, so pricing structures for this approach are usually per-session or per-seat rather than flat-fee, and availability can be constrained during peak exam windows.
AI-driven automated proctoring uses computer vision and audio analysis to monitor the candidate's webcam feed, screen, and browser activity, flagging anomalies (gaze patterns, unusual noise, a second face) for review after the fact rather than in real time. This optimizes for scale and cost efficiency, since one system can monitor thousands of concurrent sessions without a proportional increase in staff. The tradeoff is that flags require human review to resolve fairly, and candidates with certain accessibility needs, lower bandwidth, or unusual testing environments can trigger false positives that create dispute-resolution overhead.
In-person test center proctoring relies on a physical facility, either a dedicated testing center network or an on-site supervised room, where a human proctor manages the room directly and identity verification happens face-to-face. This optimizes for perceived security and defensibility in high-stakes or legally sensitive certifications, where a physical chain of custody for the exam matters. The tradeoff is accessibility and cost: candidates must travel, which adds time and expense and can suppress volume in regions with sparse center coverage, and the credentialing body typically pays facility or seat-time fees per candidate.
Hybrid dual-delivery models offer both online and in-person options for the same certification, letting candidates choose based on their own circumstances, or reserving in-person delivery for specific exam versions (like practical or performance-based components) while keeping knowledge-based sections online. This optimizes for reach without abandoning the security profile of in-person testing for the components that need it most. The tradeoff is operational complexity: running two proctoring workflows means two sets of policies, two support paths, and two data pipelines feeding into the same exam engine, which increases the coordination burden on program administrators.
How Do the Approaches Compare at a Glance?
The four models trade off differently across the variables that matter most to a credentialing body running the decision: how security is enforced, how accessible the exam is to candidates, how cost scales, and where each model tends to fit best.
| Approach | Security Mechanism | Candidate Accessibility | Cost Structure | Best Fit |
|---|---|---|---|---|
| Live remote human proctoring | Real-time human observation via video | High, no travel required | Per-session, staffing-driven | Moderate-to-high-stakes exams needing human judgment |
| AI-driven automated proctoring | Algorithmic flagging with post-hoc human review | High, but sensitive to bandwidth and device access | Usage-based, scales with volume | High-volume, lower-to-moderate-stakes certifications |
| In-person test center proctoring | Physical supervision and chain of custody | Lower, dependent on center location | Per-seat facility fee | High-stakes exams with legal or licensure weight |
| Hybrid dual-delivery model | Combination, matched to exam component | Highest, candidate or program chooses | Blended, tied to delivery mix | Programs serving a wide range of candidate stakes and geographies |
The hybrid row is the only one that avoids the security/accessibility tradeoff.
Evaluation Criteria for Credentialing Bodies
A credentialing body should revisit its proctoring criteria as the candidate population and regulatory obligations change.
Exam stakes and legal exposure: A licensure exam tied to professional practice rights (nursing, engineering, financial certification) carries more legal exposure if challenged than a vendor-neutral skills badge, and that exposure should drive how much security infrastructure is justified.
Candidate geography and device access: Programs with candidates in regions with unreliable broadband or limited access to modern webcams face higher failure and dispute rates with online proctoring, regardless of which vendor supplies it.
Accreditation and standards alignment: Bodies seeking or maintaining accreditation, such as NCCA recognition or alignment with ANSI/ISO/IEC 17024, should confirm that whichever proctoring method is chosen produces the documentation trail those standards require for exam security audits.
Data privacy and biometric regulation: Online proctoring that captures facial recognition or biometric data triggers state and national privacy law, including the Illinois Biometric Information Privacy Act (BIPA) and GDPR where candidates are in the EU. Programs should confirm data retention and consent practices before rollout, not after a complaint.
Accessibility and accommodation policy: Both online and in-person methods need a documented accommodation process for candidates with disabilities under the ADA, and the process should be tested with real candidates before it's needed in a live exam window.
Dispute and appeal process: Every proctoring method eventually flags a false positive. The evaluation should include how quickly a flagged session gets human review, what evidence the candidate can see, and how an appeal is adjudicated, since this process shapes candidate trust in the credential more than the proctoring technology itself.
What Does Implementation Involve?
Rolling out a proctoring approach, or changing an existing one, follows a sequence that most credentialing bodies underestimate on the first pass. The gating factor is rarely the technology itself; it's the policy and integration work that has to happen before any candidate sits an exam.
The first step is defining the exam security policy in writing: what constitutes a violation, what evidence is required to flag one, and who has authority to adjudicate a dispute. This has to happen before selecting a delivery mechanism, because the policy determines what the proctoring method needs to capture and retain. Skipping this step is the most common mistake in the space; organizations pick a technology first and then try to write policy around whatever the tool happens to log, which creates gaps when a dispute reaches a level where documentation matters.
The second step is integration with the exam delivery system and the learning management system that houses candidate records, eligibility data, and score reporting. Proctoring vendors typically expose an API or a scheduling handoff; the exam engine needs to pass candidate identity and eligibility data to the proctoring layer and receive session results and flags back in a format the credentialing team can act on. This integration work, not the proctoring technology's monitoring accuracy, is usually the longest part of a rollout timeline.
The third step is a pilot with a small, representative cohort before full deployment. The pilot should intentionally include candidates on lower-bandwidth connections, older devices, and any accommodation profiles the program serves, because these are the conditions that surface false positives and technical failures. Programs that skip piloting and go straight to full-scale deployment tend to discover their flag-rate and support-load problems during a live exam window, which is the worst possible time to discover them.
The fourth step is staff and candidate training. Proctors, whether human reviewers or the support staff handling AI-flagged sessions, need documented escalation criteria so that judgment calls are consistent across sessions. Candidates need clear, tested instructions for technical setup (bandwidth, browser permissions, ID requirements) delivered well before the exam date, not buried in a confirmation email. A recurring failure mode in online proctoring rollouts is technical friction discovered at the moment of the exam, which candidates experience as the exam itself failing, even when the underlying certification content and delivery are sound.
Finally, ongoing monitoring after launch matters as much as the initial rollout. Flag rates, dispute volumes, and candidate support tickets are the observable signals that reveal whether a proctoring method is working as intended. A rising flag rate concentrated among a specific device type or region is a signal to investigate the technology, not the candidates.
Frequently Asked Questions
How much does online proctoring typically cost compared to in-person testing?
Neither structure is inherently cheaper. Online proctoring avoids facility overhead but still pays for live proctors or human review of flagged sessions, while in-person testing adds the candidate's own travel cost, which the credentialing body doesn't pay directly but which affects candidate volume. Total cost depends on exam volume, geographic spread of candidates, and how much human review each flagged session requires.
Is online proctoring secure enough for high-stakes certification exams?
Yes, when it combines live human observation or AI flagging with a documented review and appeal process, online proctoring can meet the security bar for high-stakes exams, and some programs already use it this way. The determining factor isn't the technology alone but whether the credentialing body can produce an audit trail that satisfies its accreditation requirements, such as the exam security documentation called for under ANSI/ISO/IEC 17024, if a result is ever challenged.
What's a common misconception about switching from in-person to online proctoring?
The most common misconception is treating it as a one-time technology swap rather than a policy and integration project. Organizations that assume the proctoring vendor's default settings will match their existing security policy often discover gaps only after a dispute arises, because the exam security policy, the accommodation process, and the data integration with the exam engine all need to be built or rebuilt around the new delivery method, not layered on top of it.
Can a certification program run online and in-person proctoring at the same time?
Yes, and this hybrid model is common among associations and credentialing bodies serving candidates with a wide range of geography and exam stakes. Scoping each method to where it fits best, rather than running both everywhere, is what keeps the added complexity manageable.
Do candidates prefer online proctoring over in-person testing?
Preference varies by candidate circumstance rather than trending uniformly one direction. Candidates without reliable broadband, quiet testing space, or comfort with webcam monitoring often prefer in-person centers despite the travel, while candidates balancing work schedules or living far from a test center consistently favor the flexibility of online delivery. Programs that offer both and track which candidates choose which option, and why, get the clearest signal for how to weight future exam delivery investment.