The brief
An education and professional-learning organisation struggled to manage quizzes, guided exercises, self-assessments, and exam simulations. Content owners juggled manual creation and multilingual support while learners faced slow, inconsistent feedback. They needed a modern website to host it all, plus AI automation to take the manual work off content owners.
A legacy system blocking content
The organisation ran its assessment content across a mix of spreadsheets and a legacy CMS that required developer time for every update. Creating a new quiz took hours. Translating it for another language took days. Learners experienced inconsistent question formatting and had no real-time view of their own progress, which drove low completion rates and poor exam readiness.
The cost of that legacy setup was never just inconvenience. Every content change was billed at developer rates, and the queue of pending updates regularly ran three to four weeks deep. Subject matter experts would draft new assessment material, hand it off, and wait. By the time updates went live, some content was already out of date. The backlog itself became a liability, because the team stopped submitting changes they knew would sit unactioned.
What we built
We built TrainED from scratch: a fully customisable quiz engine where a single content owner creates questions with titles, hints, and difficulty ratings; Zapier syncs content updates automatically across the platform; and learners get a clean, responsive dashboard covering guided exercises, self-assessments, and timed exam simulations. The stack is React, Vite, Node, and Supabase, supporting multilingual content delivery without duplication.
Difficulty ratings were a deliberate design decision, not a cosmetic feature. Exam preparation requires learners to encounter harder material progressively, so the quiz engine sequences questions by difficulty when the content owner sets that mode. Learners attempting a high-stakes certification see foundational questions first, then progressively harder ones, which builds confidence and reduces the anxiety-driven abandonment that had been inflating dropout rates.
The timed exam simulation mode is architecturally separate from standard quiz delivery. It locks navigation, enforces strict time limits per section, presents questions in a randomised order drawn from the difficulty pool, and produces a scored result card that mirrors the format of the real exam. Learners know exactly what to expect on exam day. That fidelity was absent from the previous setup, where practice and assessment looked nothing alike.
Content sync previously meant a content owner emailing a spreadsheet to a developer, who would manually transcribe questions into the CMS, push to staging, and wait for sign-off before deploying. Zapier replaced that entire sequence. When a content owner updates the source document, Zapier triggers a webhook that writes directly to the Supabase tables the quiz engine reads from. The content is live within minutes, and the developer is no longer in the loop at all.
Multilingual delivery was architected around a locale key attached to every content record rather than separate content trees per language. A question exists once in Supabase with translations stored as structured fields on the same row. When a learner selects a language, the platform reads the appropriate field rather than fetching a different dataset. Adding a new language requires populating those fields, not rebuilding any content hierarchy, which is why rollout went from a multi-week project to a configuration change.
The content owner's workflow today is a browser-based form with a live preview. They write the question, add a hint, set the difficulty rating, attach it to one or more modules, and publish. The change is reflected for learners within minutes. Before TrainED, the same task involved a spreadsheet, an email thread, a developer, a staging review, and a deployment window. The organisation now ships content updates on the same day a subject matter expert requests them.
The outcome
New quiz setup time fell 80%, from a half-day task to under an hour. Learner completion rates rose 40% once progress tracking gave learners a clear path to the end. Content accuracy reached 95% through automated sync, eliminating the transcription errors that had been undermining exam performance. Multilingual rollout, previously a barrier, became a configuration change.