I want to be straight with you about something.
Before writing a single line of code for FailFast, I spent time studying the platforms that came before us in this market. To understand exactly what went wrong for them. Because if you are building an EdTech product in Nigeria in 2026 and you have skipped past Edukoya, uLesson, Tuteria, and the broader graveyard of well-funded ideas that did not make it, you are missing the most important context available to you.
We commissioned two research documents as part of that process. The first was a retrospective on the Nigerian EdTech market through 2026, anchored by a detailed post-mortem on Edukoya's shutdown. The second looked specifically at how EdTech platforms in Nigeria actually acquire and source question content from JAMB, WAEC, and NECO, covering the legal, technical, and economic sides of that problem.
What we found shaped almost every decision in how FailFast is built. I want to share it with you directly, because I think this community deserves to know what we know.
Part 1: Why Platforms Keep Failing
Edukoya built something technically sound and pedagogically coherent. What killed it was a market mismatch. The product assumed a connectivity and income reality that does not yet exist in Nigeria at the scale required for venture-backed growth.
The Data Cost Wall
After the Nigerian Communications Commission approved new tariffs in early 2025, mobile data prices jumped by more than 50%. The average cost of 1GB of mobile data went from roughly N287.50 in 2023 to between N475 and N637.50 by 2025/2026.
Edukoya's model ran on live video sessions between students and tutors. One hour of that consumed between 500MB and 1GB. At 2026 prices, that one hour cost nearly N500 in data alone, before any subscription fee entered the picture. For a household managing school fees, fuel, and groceries in an economy with consistent naira devaluation, that number does not fit the budget across a full term.
Budget data plans existed. Glo offered roughly 3.9GB for N1,000 at certain points. But network quality on cheaper plans varied so drastically that real-time video on those plans was effectively unusable. The gap between "technically has internet" and "has internet that supports a live class" is where Edukoya's user base lived. Any platform that does not design around that gap will eventually be broken by it.
The Free User Trap
Edukoya onboarded over 80,000 students. Converting those free users into paying subscribers was where the model collapsed. In an economy where parents are already stretched thin, a digital learning supplement sits at the bottom of the priority list. School fees come first. Exam registration fees come first. Food comes first.
The research described this as the "must-have versus nice-to-have" filter, and it is blunt but accurate. Nigerian parents will pay for outcomes with direct, visible consequences: UTME prep because a university place depends on it, IELTS prep because a visa depends on it. General homework help on a Tuesday evening, regardless of how good the product is, has never produced consistent willingness to pay at scale.
Edukoya also overhired to support its live model. High-quality tutors earned up to $133 per month, well above local rates. The cost of delivering the service exceeded what the market would pay for it. More free users did not fix that equation.
The Balance Sheet Nobody Talked About
The funding news was loud. The financials were quieter. By 2024, Edukoya's net liabilities had grown to £511,332, up from £353,881 in 2022. Liquid assets were declining. Sitting on the books across three consecutive filing years was a static long-term debt figure of approximately £1.95 million that financial analysts reviewing the UK filings could not fully account for.
The reasonable conclusion some drew was that a significant portion of that $3.5M pre-seed may have been structured as debt, rather than equity. If that was the case, a lot of that "funding" was already spoken for before the team could deploy it. The structure of your capital matters as much as the size of it.
Part 2: The Models That Survived and Why
The post-Edukoya picture in 2026 is not total collapse. Some platforms are still standing. What they share is that none of them bet on infrastructure improving fast enough to support their growth targets.
uLesson: Build Around the Constraint
uLesson launched with SD cards and USB drives containing pre-recorded video lessons. No stream, no live session, no recurring data cost for the user. Students watched on their devices offline.
That "offline-first" decision, which looked like a limitation at launch, turned out to be the differentiator. When data prices spiked and Edukoya's model cracked under the pressure, uLesson's model was already designed around that pressure not existing. It eventually introduced streaming for middle-class users as data access improved slightly for that segment, but the foundation never depended on it.
By 2025/2026, uLesson had also expanded into higher education through Miva Open University, securing a license to offer full degrees. That move targeted a demographic with a demonstrably higher willingness to pay: students seeking credentials for employment or migration, rather than supplementary homework help.
Tuteria: Stay Lean, Chase High Stakes
Tuteria, founded in 2015, survived by operating as a marketplace rather than a content platform. It connected students with independent tutors and took a 15 to 30 percent commission on bookings. No employed tutors, no content production costs, no live infrastructure to maintain.
When the market contracted, it could contract with it without shutting down. By 2026, Tuteria raised $2.6 million in new funding by repositioning around high-stakes exams: UTME, IELTS, GMAT. These are exams with clear, urgent outcomes. That urgency produces a willingness to pay that general supplementary learning has never matched.
AltSchool Africa: Tie Revenue to the Outcome
AltSchool's model ties payment to employment through an Income Share Agreement. Students pay tuition only after securing a job in their field. That contract ties the platform's survival to the student's actual economic outcome, which produces a very different kind of user commitment than a monthly subscription.
The value proposition is also in a different category from K-12 supplementary learning. AltSchool positions itself as an alternative to a university degree, with diplomas in software engineering and data science. The youth unemployment crisis in Nigeria gives that pitch a direct and concrete argument. The return on investment is measurable and near-term, which makes monetization far easier.
The pattern across all three is the same: build for the market as it is, not as it should be.
Part 3: The Content Sourcing Problem Nobody Talks About Enough
The second research document covers something less visible but equally important for any platform touching Nigerian exam content: where questions actually come from, who legally owns them, and where the risks sit.
The Good News: An API Ecosystem Exists
Nigeria's EdTech content infrastructure has moved further than most people realize. The MyQuest API gives developers direct access to JAMB and WAEC question banks. The Aloc.ng QBoard API is built specifically for gamified learning, structured around exam question sets. The Nigeria Past Questions API on GitHub is building an open-source database of exam content with offline-first support. In August 2025, the Henotace AI SDK launched as Africa's first AI education copilot, adding predictive analytics on top of the question pipeline.
The technical foundation for building a question-bank-backed platform in Nigeria is real and accessible. That matters for FailFast because our entire product runs on questions.
The Shadow Market Problem
Alongside the official ecosystem, a parallel market operates. Platforms exist that offer "100% verified WAEC and NECO questions" before exam dates, a practice locally called "runz." These have no formal relationship with any examination body. They source questions through unofficial channels and distribute them freely.
The risk this creates for any platform goes beyond the legal exposure. If your question bank mixes verified official content with material that originated in a leak channel, even unknowingly, you compromise the trust of every student and school on your platform.
The AI Authorship Gap in Nigerian Law
This one matters specifically to platforms using AI to generate content, which FailFast does. Nigeria's copyright framework was updated in December 2025 with a new IP Policy aimed at the digital startup ecosystem. The update left the central question open: who owns content that an AI generates?
Nigeria's copyright law is built on the assumption of human authorship. When a generative model produces a practice question from a curriculum prompt, there is no statutory answer to who holds that copyright. Jurisdictions like the UK, India, and Hong Kong have amended their laws to accommodate this. Nigeria has not yet. For any platform building a proprietary question library using AI, that gap is a documented business risk. We know it exists and we are operating with that knowledge in view.
Part 4: Where FailFast Sits in All of This
Everything above was known before we started building. These were inputs, not surprises. Here is exactly how they shaped our decisions.
Text Over Video, Always
FailFast is a text-based question-and-answer platform. No live sessions, no video streams, no synchronous delivery of any kind. A full FailFast session, questions, answer options, feedback, scoring, uses a fraction of the data a single minute of video requires. A user on a N1,000 Glo data bundle can practice for hours without worrying about it running out.
That was a deliberate choice. The learning model we use, multiple-choice questions with up to three attempts per question and immediate feedback, delivers full value on cheap data plans without any degradation.
Exam Content, Tied to Real Consequences
FailFast Arena's question bank draws from WAEC, JAMB, NECO, and NERDC-aligned secondary school curriculum content. Mathematics, English, Biology, Chemistry, Physics, Government, the subjects that determine whether a student gets into university.
FailFast is practice for the exams that carry real consequences. That puts us in the "must-have" category from the start, which is the only category in Nigerian EdTech where willingness to pay has proven consistent.
For the Arena user who is outside the student demographic, the competitive staking layer adds a different motivation entirely: the chance to earn real money through knowledge. That is a separately motivated user group, and having both means the platform is not dependent on a single reason to exist.
Revenue From Day One, Not Month Seven Hundred
The freemium trap that caught Edukoya required converting a massive free base into paying users, with no alternative revenue path. FailFast's model works differently.
FailFast EDU, the B2B product for schools and tutorial centers, runs on subscriptions from the first paid cohort. The first five partner schools get six months free to reduce adoption friction, but the model is subscription-based from the point those six months end and the pipeline is already in motion.
FailFast Arena's staking revenue is gated behind our gaming license, which we expect around month seven of operations. That delay is planned for in the financial model. School subscriptions and sponsored competition fees carry the first six months entirely.
A Question Pipeline With a Clear Quality Gate
For content sourcing, we draw from three verified API sources: MyQuest for JAMB and WAEC content, Aloc.ng QBoard, and the Nigeria Past Questions API. AI-generated questions supplement the bank using prompts aligned to the NERDC 2025/2026 syllabus, with a minimum of 5,000 verified questions at launch.
Every question in the system carries a status. Unverified questions, including AI-generated and community-submitted content, are available for practice in the FailFast Mine but cannot appear in real-money staking competitions. Verified questions have passed AI-assisted review for formatting, accuracy, and duplicate detection, plus manual checks on top. The integrity of paid competitions does not depend on the pace of verification.
On the AI authorship question, we treat all AI-generated questions as platform-owned for operational purposes, while the legal framework catches up. The gap is documented, not ignored.
Skill as the Foundation, Verified by the Record
The regulatory question about platforms mixing knowledge and money always comes back to one point: is this skill or chance?
FailFast's tier system answers that structurally. No user can access real-money staking without first earning 5,000 success points through verified practice. Success points only come from answering questions correctly. There is no path to the staking threshold through failure alone. The higher the tier, the harder the questions required and the higher the stake range available.
A Grand Master tier user has, by documented track record, correctly answered a large number of university-level questions before accessing the top stake range. That is a verifiable history built question by question in practice mode.
The Challenges We Are Still Carrying
Designing around known problems does not make them disappear.
Data costs are still a real pressure on our potential user base, especially at the lower income levels we want to reach through FailFast EDU's school partnerships. We designed for low bandwidth, but a student without a data plan at all cannot use a mobile app regardless of how light it is.
The gaming license timeline is real. Staking revenue does not exist until that license is in place. Our financial model accounts for that gap, but it means we are running a more constrained operation in the first six months than we would prefer.
The AI content ownership question remains unresolved in Nigerian law. We have a documented position on it and we are watching the regulatory conversation, but watching is different from having clarity.
And whether FailFast's fail-forward model actually produces better learning outcomes than conventional practice platforms is something the beta will tell us. We built the scoring system around that belief. Belief is not data, and we know the difference.
We are launching because we have spent enough time on the questions to know which risks are worth taking.
FailFast Arena's beta opens with gift-card-based staking for a controlled group of early users. FailFast EDU school onboarding runs in parallel. If you want to be part of either, follow us on X (Twitter) at @failFastNG.