Aviation Infinity at 25,000 Students: Scaling an EdTech Product

This month, Aviation Infinity crossed 25,000 registered student pilots. For a niche educational platform built and run by a solo founder, that number feels significant. It's also a good moment to reflect on what I've learned about scaling an EdTech product, because the challenges at each stage were very different from what I expected.

The First 1,000

The first thousand users were the hardest to get and the easiest to serve. I found them through aviation forums, student pilot groups, and direct outreach to flight schools. Each signup felt like a personal victory. I knew many of them by name. When they had issues, they'd email me directly and I'd fix things within hours.

The product at this stage was simple: a question bank with explanations. Students could practice exam questions, see their score, and read explanations for questions they got wrong. There was no progress tracking, no study plans, no adaptive learning. Just questions and answers.

And it was enough. The first thousand users validated that online exam preparation was something student pilots wanted. They didn't need a sophisticated product. They needed a reliable one with accurate content.

The lesson from this stage was that initial product-market fit in EdTech is about content quality, not feature sophistication. If your content helps students learn, they'll tolerate a basic interface. If your content is mediocre, no amount of gamification or UX polish will save you.

1,000 to 5,000: The Quality Crisis

Somewhere around 2,000 users, I started getting emails about question accuracy. Not many, maybe one or two per week. But each one was serious. A student would find a question where the "correct" answer was wrong, or where the explanation contradicted the official study material.

In an exam prep product, content accuracy isn't a nice-to-have. It's existential. If students can't trust that the questions and answers are correct, the entire product is worthless. Worse than worthless, because incorrect practice material can cause students to fail their actual exams.

I spent three months building a content quality system. Every question went through a verification pipeline: initial creation, expert review, cross-referencing with official documentation, and user feedback integration. I also built a reporting system that let users flag questions they believed were incorrect, with each flag triggering a review.

This was boring, unglamorous work. No user ever signed up because of the content verification pipeline. But it was the foundation that everything else was built on. Without trusted content, nothing else mattered.

5,000 to 10,000: The Feature Demands

Once the content was trustworthy and the user base was growing, the feature requests started arriving. Students wanted progress tracking. They wanted to know which topics they were strong in and which needed more work. They wanted study plans that told them what to study and in what order. They wanted mock exams that simulated the real exam experience.

This stage was a product management challenge. I was a solo founder with limited development time, and the feature requests were endless. I had to be ruthless about prioritization.

The framework I used was simple: which feature will help the most students pass their exams? Not which feature will attract new users, not which feature is technically interesting, but which feature will directly improve exam outcomes for current users.

This led me to build adaptive question selection before anything else. Instead of random question sets, the platform started identifying each student's weak areas and prioritizing questions in those areas. The impact on student performance was immediate and measurable: average practice scores improved within weeks of the feature launch.

Progress tracking came next, followed by structured study plans. Mock exams came last, even though they were the most requested feature, because they helped students assess their readiness but didn't directly help them learn.

10,000 to 25,000: The Scale Challenges

The jump from 10,000 to 25,000 users introduced challenges that were fundamentally different from the earlier stages.

Performance became a real concern. With thousands of students practicing simultaneously, database queries that had been fast enough at smaller scale started showing latency. The question recommendation algorithm, which worked by analyzing a student's entire answer history, became expensive to compute as some students accumulated thousands of answers.

I rewrote the recommendation engine to use pre-computed strength scores that updated incrementally with each answer, rather than recomputing from scratch on every session. I added caching for frequently accessed content (popular question sets, explanation pages) that reduced database load significantly. I optimized the most common query patterns with targeted indexes.

These weren't exciting engineering challenges. They were the mundane work of making a system perform well at scale. But they were critical. A platform that's slow during peak study hours (evenings and weekends) loses students to frustration.

Customer support scaled in a way I wasn't prepared for. At 5,000 users, I could handle support emails personally. At 25,000 users, that was impossible. I built a comprehensive FAQ system, added in-app guidance for common workflows, and created help documentation that addressed the most frequent questions. This reduced support volume enough that I could continue handling the remaining requests personally, but it was a close call.

The Monetization Evolution

Aviation Infinity's business model evolved significantly as the user base grew. The early product was free with optional donations. This was naive but got me to the first few thousand users.

At around 3,000 users, I introduced a freemium model: basic access was free, and full question bank access required a subscription. This worked but left revenue on the table because the free tier was generous enough that many students never needed to upgrade.

At 10,000 users, I restructured pricing with the experiments I described in my earlier post about pricing. Higher prices, annual plans, and a more limited free tier significantly improved revenue per user.

The current model at 25,000 users is a straightforward subscription with monthly and annual options. There's a free trial that gives full access for a limited time. The pricing reflects the genuine value of the product: helping students pass exams that cost hundreds of euros in fees and thousands in training time.

What EdTech Gets Wrong

Building Aviation Infinity has given me opinions about what the EdTech industry gets wrong.

Too much focus on engagement metrics and not enough on learning outcomes. Gamification, streaks, badges, and leaderboards are engagement tools, not learning tools. They keep users opening the app but don't necessarily help them learn. I've resisted adding these features to Aviation Infinity because I want students to study because the platform helps them learn, not because they're chasing a streak.

Over-reliance on video content. Video is great for some types of learning, but for exam preparation, interactive practice with immediate feedback is dramatically more effective. Students who actively answer questions and review explanations retain more than students who passively watch video lectures. Our data confirms this consistently.

Ignoring the instructor relationship. EdTech platforms often position themselves as instructor replacements. This is a mistake in aviation education and probably in most fields. The best learning outcomes happen when platform and instructor complement each other. Aviation Infinity works best when a flight school instructor uses it to assign practice and track student progress, not when a student uses it in isolation.

Looking Ahead

At 25,000 students, Aviation Infinity is the largest independent pilot exam preparation platform in its market segment. That's gratifying, but it's also a responsibility. Thousands of students are relying on this platform to help them pass exams that determine their careers.

The roadmap from here focuses on three areas. Deeper learning analytics that give students and instructors clearer insights into knowledge gaps. Expanded content coverage for additional exam types and regulatory frameworks. And improved mobile experience for students who study primarily on their phones.

Twenty-five thousand students is a milestone, not a destination. The aviation training market is large, and there are student pilots in dozens of countries who could benefit from quality exam preparation. The next milestone is somewhere around 50,000, and getting there will require solving a new set of challenges that I can't fully anticipate yet.

That's what makes building products exciting. Every scale milestone reveals the next set of problems. You're never done.