Observer is a groundbreaking system – we hope you have more questions!
We sat down with the Observer team to compile a list of the most common questions they hear about Observer's features, capabilities, and security advantages. Read on to learn more.
Observer is a data-driven test administration monitoring solution that evaluates live exam data in real-time to identify potential security risks. Unlike traditional proctoring systems that rely solely on human or AI surveillance of a test taker, Observer analyzes test-session data to detect unusual patterns and alert trained human reviewers when needed.
Traditional proctoring treats every test session as an equal risk and relies solely on fallible, continuous human or AI surveillance. Observer uses real-time data analysis to assess risk objectively and calls in trained reviewers when needed. By focusing on data instead of people, Observer offers a respectful, secure, and cost-efficient alternative to conventional proctoring software.
Observer continuously collects and analyzes data during an exam, including test-design elements, real-time test session data, and typical proctoring cues. Its patented risk analysis system interprets these data streams to flag potential risks and notifies a trained reviewer if attention is required.
Testing programs adopt Observer to achieve an improved testing experience, better exam security, and dramatic cost savings compared with traditional or AI-based proctoring systems.
Observer Plus combines Observer’s real-time monitoring with secure test design to deliver the gold standard in online exam security. This combination prevents pre-knowledge and neutralizes content theft while addressing the security concerns already managed by Observer.
While Observer focuses on monitoring data from live test sessions, Observer Plus adds secure test and item design to make stolen content and pre-knowledge useless. This layered approach addresses the most common and damaging security threats that monitoring systems simply cannot detect or prevent.
Caveon is the world’s leading test security company. For more than 25 years, our mission has been to protect the integrity of exams. One of the greatest challenges to that mission has been the over-reliance on traditional online proctoring, which is expensive, invasive, and ultimately ineffective at providing true security.
Observer is our response to this broken system. It was designed by experts who understand both the science of test security and the human side of testing. Our mission has never been about quick profits or hype. It’s about setting a higher standard for integrity in every assessment.
The Risk Analysis Management System (RAMS) is the patented analytics engine that powers Observer. It analyzes multiple data streams (test design, session behavior, and proctoring cues) to detect potential security risks in real time.
Observer’s patented risk analysis system, RAMS, evaluates multiple streams of data during every test session to identify potential security risks. These can include:
Test Design Data — Information about test structure, item pools, question exposure, and delivery method.
Real-Time Session Data — Live exam session data such as response times, navigation, and browser activity.
Traditional Proctoring Cues — Video, audio, and screen-share inputs
Learn more about Observer’s data-driven approach and the RAMS engine on our How Observer Works page.
Caveon continually enhances the RAMS engine to improve its accuracy, adaptability, and performance. As testing data grows and new inputs are incorporated, RAMS will further refine its ability to identify emerging exam security threats.
Caveon offers a complete, end-to-end testing solution built through seamless integrations with trusted partners. Our ecosystem connects best-in-class tools for learning management, candidate management, voucher management, delivery, badging, and more.
Contact us if you're interested in hearing more about our all-inclusive offerings.
No. Observer doesn’t rely on restrictive lockdown browsers. Instead, it monitors browser activity to detect and flag suspicious actions like opening new tabs, leaving the exam window, or using copy/paste commands.
Observer will detect and pause an exam if AI extensions, also known as “homework helpers," are present.
No software can guarantee that an exam isn’t being captured by readily purchasable hidden camera devices. That’s why Caveon recommends combining Observer with Secure Test Design through Observer Plus. By embedding security directly into the exam, stolen or recorded content becomes useless.
Observer verifies test-taker identity through photo and ID comparison, with optional integration for 3rd party identity verification tools (eg, CLEAR or ID.me.)
Observer is designed to use Caveon’s patented proxy-detection method to identify when a different person is taking the test. This feature is coming soon (as of October 25, 2025).
Observer Plus is the solution to pre-knowledge and content theft. Secure test design ensures that no two examinees receive the same combination of questions, making stolen or shared content useless.
Caveon’s test development experts can help you evaluate your current exam structure and convert it into a secure, data-driven design for Observer Plus.
Each testing program defines its own security thresholds and determines which types of activity should trigger a review, within the limits of Observer’s available configuration options.
These settings can also vary by exam type. For example, a high-stakes exam might use stricter oversight, while a practice test might use lighter settings.
Observer includes pre-set security configurations for various exam types (eg, high stakes, practice, classroom). These settings provide a strong starting point and can be configured to reflect your program’s specific goals and testing environment.
A trained human reviewer is automatically notified and assigned a short, specific review task based on the data indicators.
The reviewer examines only the relevant portions of the session to determine whether the event represents a security concern. If intervention is necessary, they follow pre-established policies (such as pausing or ending the exam). If not, the test continues seamlessly, and the test taker remains undisturbed.
Programs can use Caveon’s trained observers or staff their own teams under the Bring-Your-Own-Personnel (BYOP) model for cost-effective monitoring.
The number of observers depends on a variety of factors. In practice, two observers have been able to monitor up to 100 test sessions at once, with average review response times under two minutes once alerted.
Observer supports the current versions of Chrome, Firefox, Safari, and Edge on Windows, macOS, Linux, and Chromebooks.
No. Observer runs entirely in the browser. There are no downloads, extensions, or extra accounts needed.
The recommended minimum connection speed is 20 Mbps download / 10 Mbps upload. Stable internet connectivity is required.
If the connection is interrupted, the candidate is returned to the setup screen. When connectivity and required inputs are restored, the exam resumes from where it left off.
Yes. Observer is fully LTI 1.3 compliant and integrates seamlessly with leading LMS and CMS, platforms including Docebo, Skilljar, Moodle, Canvas, Blackboard, and Edmentum. If you don’t see your platform listed, contact us to discuss an integration.
No. Observer integrates with most testing platforms via LTI 1.3. Some features currently work only with Caveon Scorpion, but custom integrations can be configured.
Observer supports LTI 1.3 and can integrate with your existing testing systems. Caveon’s integration team is available to assist with setup.
AI helps identify potential anomalies (e.g., additional people, unexpected noises, unusual behavior) and surfaces flags for human review. AI outputs do not make decisions about examinees; all exam‑impacting judgments require trained human review.
No. Human oversight is required. Clients must not use Observer to make automated, determinative, or legally binding decisions about examinees without independent human review.
No. Observer does not use personal data to train its AI models. De‑identified session and chat data may be retained to monitor, evaluate, and improve AI components and support processes.
Caveon plans to obtain ISO/IEC 42001:2023 certification and publish the associated Statement of Applicability by June 2026.
Observer complies with major data protection standards as specified within frameworks of SOC 2, ISO 27001:2022, and ISO 42001:2023 Caveon conducts regular, annual audits to verify that these standards are met and to ensure that all test taker data is managed responsibly. To date Caveon is SOC2 Type II audited, and ISO 27001:2013/2023 certified and is actively seeking ISO 42001:2023 certification.
No. Observer does not create or use facial templates/embeddings or voiceprints and does not otherwise process biometric identifiers.
Observer is being developed to align with WCAG 2.2 (AA for examinees; A‑level for observers). An Accessibility Conformance Report (VPAT) will be provided once testing and validation are complete.
