How is this different from the quizzes in a learning management system?
By stakes and by what failure costs. An LMS quiz supports learning: it is formative, low-consequence, and a technical fault is an inconvenience. A formal examination is a high-stakes event where a failed submission, a leaked paper, or an unfair advantage can invalidate a result and expose the institution to appeal. That difference drives real requirements an LMS quiz engine does not have: item banking with security and statistics, examination sessions with candidate identity assurance, proctoring, resilient submission under load, marking with moderation and second-marker workflow, and defensible audit. Continuous assessment belongs in the LMS; examinations belong in a platform built for them.
What is item banking and why does it matter?
An item bank is the institution's assessment asset: questions stored with their metadata rather than embedded in individual papers. Each item carries the topic or learning outcome it tests, difficulty, historical performance statistics, its author and review status, and its exposure history. That structure enables the things institutions actually need: generating equivalent papers from a blueprint rather than writing each one from nothing, retiring items whose statistics show they are ambiguous or over-exposed, and demonstrating that a paper covers the syllabus as claimed. Without item banking, an assessment platform is a form builder and every examination session starts from scratch.
Do you build proctoring, or integrate it?
We integrate it, and we are direct about its limits. Automated proctoring, whether by camera, screen monitoring, or behavioural analysis, is provided by specialist vendors and is not something an education project should build. More importantly, no proctoring technology detects cheating; it produces signals that a human must interpret. We integrate the provider your policy selects, surface flagged sessions for human review with the evidence attached, and design the process so a flag is never an automatic accusation. We also advise on assessment design that reduces the need for surveillance, since randomised item selection, time-constrained delivery, and questions requiring applied reasoning are frequently more effective than monitoring.
How do you make sure submissions are not lost?
By treating this as the platform's primary obligation, because a lost examination submission is an academic incident rather than a bug. Candidate work is saved continuously as it is produced rather than only on submission, held locally as well as on the server so a network interruption does not lose keystrokes, submitted with an explicit acknowledgement the candidate can see, and reconciled at session close so any candidate whose work did not arrive is identified immediately rather than at marking. We load-test against real examination profiles, where hundreds or thousands of candidates start within the same minute and submit within the same minute, which is a far harsher pattern than ordinary application traffic.
How are accommodations and accessibility handled?
As a core function rather than a special case, because in many jurisdictions accommodations are a legal entitlement and getting them wrong invalidates the assessment. The platform must support additional time as a per-candidate multiplier or fixed extension, rest breaks that pause the clock, alternative formats including screen reader compatible delivery and larger text, assistive technology permitted inside a locked-down environment, scribe and reader arrangements, and separate or supervised delivery. Accommodations flow from the student record rather than being configured per examination by hand, since manual configuration is where entitlements get missed. We build to WCAG 2.1 AA and verify with assistive technology; we do not claim conformance certification, which is a property of a verified build rather than a vendor.
How does marking and moderation work?
Objective items mark automatically; everything valuable about the platform concerns the rest. Written responses need allocation to markers with workload balance, anonymous marking where policy requires the candidate's identity withheld, rubric-based scoring with criterion-level comments, second marking and blind double marking, reconciliation where markers disagree beyond a tolerance, moderation sampling across a cohort, and a board process that can adjust at cohort level with the adjustment recorded rather than silently applied. Every score change retains its prior value, its reason, and its approver, because appeals are decided on that trail.
How is examination security handled before and during the session?
Paper security is a lifecycle problem, not a delivery feature. Before the session, that means authoring access restricted and audited, papers encrypted at rest with keys released on a schedule, no complete paper visible to any single unauthorised role, and versioned approval before publication. During the session, it means candidate authentication appropriate to the stakes, item and option randomisation to limit collusion value, time control that survives a page reload, browser lock-down integration where policy requires it, and detection of concurrent sessions on the same account. After the session, it means item exposure tracking so an over-used item is retired, and a full audit of who accessed what and when.
What analytics does an assessment platform need?
Two distinct kinds, for two audiences. Psychometric analysis serves assessment quality: item difficulty and discrimination, distractor performance, reliability measures across a paper, and flags for items that behaved unexpectedly and may need review before results are released. Educational analysis serves teaching: performance by learning outcome rather than by question, cohort comparison over time, and identification of syllabus areas where a whole group underperformed. Institutions frequently ask for the second and benefit most from the first, because item-level statistics are what make each subsequent examination cycle better than the last.
Can it handle remote and on-site examinations together?
It has to, since most institutions now run both and often within the same cohort. The platform models the delivery mode as a property of the session rather than of the system: an on-site session in a controlled room with institutional devices and invigilator sign-in, and a remote session with proctoring integration and candidate device checks, both drawing from the same item bank and both producing the same result record. The difference lives in identity assurance, environment control, and network assumptions, not in the assessment itself. Designing for one mode and retrofitting the other is a common and expensive mistake.
How long does an assessment platform project take?
A focused first release covering item banking, paper authoring, delivery with resilient submission, automatic marking, and manual marking with moderation is typically five to nine months including integration and a live pilot examination. The schedule is shaped less by feature count than by two factors: how much variation exists in your marking and moderation regulations, and whether proctoring and lock-down requirements apply. We insist on piloting with a real but low-stakes examination before any high-stakes session, because a platform's first genuine test is a full cohort submitting at once.
How do you handle candidate data and defensibility?
Assessment records are appealed, audited, and sometimes litigated years later, so the platform must be able to reconstruct any candidate's examination exactly: the paper they received, when they started and finished, what they submitted and when it was acknowledged, who marked it against which rubric version, every score change with its reason and approver, and any accommodation applied. We build immutable audit on all of that, role-based access separating authoring, invigilation, marking, and administration, encryption in transit and at rest, and defined retention per record class. Our information security management is certified to ISO 27001:2013 and quality management to ISO 9001:2015. We do not claim accreditation with any examination or education regulator; we build to the standards you are obliged to meet under your own governance.
