From Past Papers to Question Bank: AI-Powered CBT Setup
Most schools across West Africa have filing cabinets and hard drives full of past papers. WAEC questions from 2015 to 2024. NECO past questions printed and bundled by subject. Internal mock exams from five academic sessions ago. School coordinators know these materials are valuable — students who practise with past questions consistently perform better — but turning those papers into a usable, searchable, assignable question bank has always been the obstacle.
The traditional process is manual and slow: someone reads each question, types it out, identifies the options, marks the correct answer, tags the topic, and loads it into the system one by one. For a school with fifty past questions per subject across ten subjects, that is five hundred individual data entry tasks before the first CBT exam can run.
Femlify's AI changes that equation entirely. Upload a past paper — PDF, Word document, or plain text — and the AI parses every question, detects the four options, identifies the correct answer, and loads the results directly into your CBT question bank. What previously took a staff member two or three days can now be done in minutes.
This post walks through exactly how that works, and how Nigerian schools can go from a pile of past papers to a live, running CBT exam faster than most ICT coordinators think is possible.
What a CBT Question Bank Is — and Why It Changes Exam Management
A CBT (Computer Based Testing) question bank is a centralised, searchable library of exam questions that teachers and administrators can draw from when creating tests and examinations. Instead of writing a fresh set of questions for every exam, a question bank lets you assemble an exam by selecting from questions that are already written, vetted, and tagged by subject, topic, and difficulty.
The practical advantages for a Nigerian school are significant:
Consistency across sittings. When questions come from a shared, reviewed bank, the difficulty and scope of exams becomes more predictable and fair — particularly important for JS and SS classes where term scores contribute to continuous assessment.
Exam variety without extra work. A large question bank allows different students to receive different questions in the same sitting — reducing the opportunity for exam malpractice without requiring teachers to write multiple paper variants manually.
WAEC and JAMB readiness built in. When your question bank is populated with actual WAEC past questions organised by topic, every internal CBT becomes a form of structured exam practice. Students are not just sitting a school exam — they are building familiarity with the exact format, phrasing, and difficulty profile they will face in the national examination.
A well-built CBT question bank doesn't just make exam administration easier — it makes every internal exam a rehearsal for the one that matters most.
The Traditional Way Schools Set Up CBT (And Why It Stalls)
For most Nigerian schools that have attempted to set up a CBT system, the question bank is where the process breaks down.
The technical infrastructure — computers or tablets, an exam platform, a network — gets purchased and installed. The ICT coordinator is trained. The system is ready. And then the question bank sits empty because no one has had time to populate it.
Here is what the manual process typically looks like:
A teacher or administrator opens a past paper. They type question one into the system — copying the stem, the four options, and flagging the correct answer. They tag it by subject. If they are thorough, they also add a topic tag and a difficulty rating. Then they move to question two. For a 60-question WAEC past paper, this takes between 90 minutes and three hours depending on how carefully it is done — and that is for a single paper.
Across ten subjects, three years of past questions, and JS and SS levels, the backlog quickly becomes months of work. Schools that start the process rarely finish it. Those that finish it often produce question banks full of typos, inconsistent tagging, and missing correct-answer flags — because manual data entry at volume degrades in quality.
The result: the CBT system exists but is never fully used. Exams still get printed on paper. The investment in hardware sits idle. And students miss out on the practice benefit.
The question bank is where CBT ambitions go to die in most Nigerian schools — and it doesn't have to be that way.
How Femlify's AI Converts Past Papers into Structured Questions
Femlify's AI Question Extraction tool was built specifically to solve this problem. The process is straightforward enough that an ICT coordinator with no AI experience can run it on their first attempt.
Step 1: Upload the source document. Femlify accepts PDF files, Word documents, and plain text. A WAEC Biology past paper saved as a PDF, a NECO Mathematics paper exported from a scanned document, an internal mock exam typed in Word — all work as inputs.
Step 2: The AI parses the document. Femlify's AI reads the document and identifies the question structure automatically. It detects question numbers, reads the stem of each question, identifies the multiple-choice options (whether they are labelled A–D, a–d, or numbered), and determines the correct answer where it is indicated — either in the document itself or in a separate answer key.
Step 3: Review the extracted questions. The AI presents the extracted questions in a structured list before loading them into the bank. The ICT coordinator or subject teacher can scan the output, correct any parsing errors — an unusual formatting choice in the original document can occasionally trip the AI — and confirm before committing.
Step 4: Questions load into the CBT bank. Once confirmed, the questions are loaded directly into Femlify's CBT question bank, tagged with the source subject and ready for further organisation.
The AI does not replace teacher judgment — it eliminates the data entry work so teacher judgment can be applied to reviewing content rather than transcribing it.
For schools working through WAEC past questions from multiple years, the throughput difference is dramatic. A coordinator can process ten years of Biology past questions — roughly 600 questions — in the time it would previously have taken to manually enter one year's worth.
Organising Questions by Subject, Topic, and Difficulty
Raw extraction is only the first step. The real power of a question bank comes from its organisation — and Femlify makes it straightforward to tag and sort questions after extraction.
By subject: Questions are assigned to a subject during the extraction process. A coordinator uploading a Chemistry paper assigns it to Chemistry; the questions inherit that tag automatically.
By topic: This is where teacher input adds the most value. After extraction, a subject teacher reviews the questions and applies topic tags — for example, "Organic Chemistry," "Acids and Bases," or "Electrolysis" for Chemistry questions. This investment pays off every time an exam is assembled: instead of scrolling through 400 Chemistry questions, a teacher creating a mid-term test on Organic Chemistry filters to that topic and selects from 60 targeted questions.
By difficulty: For schools building JAMB readiness programmes or differentiated assessments, difficulty tagging allows exam creators to build tests with a deliberate spread — a 40-question exam might draw 15 questions from the "Foundation" tier, 20 from "Intermediate," and 5 from "Challenge," producing a bell-curved difficulty profile rather than a randomly weighted one.
For SS2 classes specifically — where WAEC preparation is the central academic priority — this level of organisation means that every CBT exam from second term onwards can be built from real WAEC-style questions, tagged by the exact topics on the WAEC syllabus, at difficulty levels that progressively build student confidence and capability.
Organisation turns a question bank from a storage system into a teaching instrument.
How Teachers Review and Approve AI-Generated Questions
One concern some academic heads raise about AI-generated content is quality control: what if the AI misreads a question or flags the wrong answer?
Femlify's workflow is designed with this concern in mind. The AI produces a draft extraction — not a final, published set of questions. Every question passes through a teacher review step before it enters the active question bank.
In practice, the review process is far faster than starting from scratch. A Biology teacher reviewing 60 AI-extracted questions from a WAEC past paper is checking for accuracy, not transcribing. Most questions will be correct on the first extraction. The teacher's attention is drawn to edge cases — questions where the formatting in the original document was ambiguous, where a diagram is referenced but not extractable, or where the correct answer marking was in a separate column that the AI interpreted differently.
These corrections take seconds each. The teacher makes the edit inline, confirms the correct answer, and moves on. A review of 60 questions that would have taken three hours to enter manually can be reviewed for accuracy in under 30 minutes.
Once reviewed and approved, questions are live in the bank. They can be used immediately in any exam created in Femlify's CBT module, by any teacher in that subject, for any class.
A teacher reviewing AI-extracted questions is doing the high-value work — quality judgment — while the AI has already done the low-value work — transcription.
Running Your First CBT Exam After Setup
With a populated and organised question bank, setting up a CBT exam in Femlify takes minutes rather than days.
The exam creator selects the subject, chooses questions from the bank — either manually picking specific questions or using a filter to draw a random selection from a topic and difficulty range — sets the duration, assigns it to a class, and publishes. Students access the exam through their Femlify portal on the day, work through the questions on screen, submit, and receive their score.
For the first sitting, a few practical steps make the experience smoother:
Run a pilot with one class first. Before rolling CBT out school-wide, run a single exam with one SS2 or JS3 class. This lets the ICT coordinator identify any device or connectivity issues in a low-stakes setting.
Brief students on the interface. Nigerian students who have not used CBT before will need a short walkthrough — how to flag a question for review, how to navigate between questions, and how to confirm their final submission. Five minutes before the exam is enough.
Use the results data. Femlify captures per-question performance data after a CBT sitting. If 80% of students got question 14 wrong, that is a teaching signal — the topic needs revisiting. This is one of the most underused advantages of CBT over paper exams: the data tells you exactly where student understanding broke down.
For schools building JAMB readiness, the end state is a school where every internal exam from JSS 1 through SS3 is a CBT sitting, every sitting draws from a curated question bank of WAEC and NECO-standard questions, and every result feeds back into a picture of where each student stands relative to where they need to be.
Femlify makes that system buildable — starting with a single uploaded past paper.
Frequently Asked Questions
How accurate is Femlify's AI at extracting questions from past papers?
Accuracy is high for standard past paper formats where questions are clearly numbered and options are labelled A–D. The AI performs best on clean PDF or Word documents. Scanned or photographed papers may produce lower accuracy depending on scan quality. Femlify always presents extracted questions for teacher review before they enter the bank, so any errors are caught before they go live.
Can we use questions from WAEC and NECO past papers legally in our school's question bank?
Using WAEC and NECO past questions for internal school practice and examination preparation is standard practice across Nigerian schools and is widely accepted as legitimate educational use. Your school's CBT question bank is a private internal system, not a published product. If you have specific concerns about your institution's policy, consult your academic coordinator or legal adviser.
What happens if the AI extracts a question incorrectly — for example, marks the wrong answer as correct?
Femlify's workflow requires teacher review of all AI-extracted questions before they are committed to the active question bank. Any errors — wrong correct-answer flag, misread option, or truncated question stem — can be corrected inline during the review step. No AI-extracted question enters a live exam without passing through a human approval step first.
