Six stages take a script from the scanner to a published result. Each one has a clear owner and hands over to the next automatically.
Scanning centres upload answer books as they are scanned. Uploads are chunked and resumable, file types are checked by content, and every file receives a checksum so nothing can be swapped later.
Each page's barcode or QR code is read to group pages into one script per candidate and put them in order. Pages that cannot be placed go to an exception queue where an administrator can drag them into the right script.
Every script is given a random code before anyone marks it. Scripts, or individual questions, are then distributed by subject, language and workload, with conflict-of-interest rules that keep examiners away from their own centres.
Examiners open the next script and see the question, rubric and model answer beside the page viewer. They annotate, score each parameter and submit. Work autosaves, and the platform can require every page to be viewed before submission.
A configurable share of scripts is marked twice. When the two totals differ by more than your threshold a third evaluation is opened automatically. Control scripts with known scores measure each examiner's accuracy, and re-evaluation requests follow a tracked workflow.
Objective scores from OMR or computer-based tests are imported and merged with subjective totals. Rounding, grace and moderation rules are applied, a validation report is reviewed, and results are published in controlled batches.
Book a walkthrough. Bring a sample paper and we will show it moving from scan to result.