I still remember the day we hit 1,000 users. We celebrated with cheap pizza and a sense of invincibility. We thought we had it all figured out. A few years and 49,000 more users later, I can tell you we knew almost nothing. Growing a learning platform from a garage project to a tool used by a small city’s worth of students taught me more than my engineering degree and two exits combined. It’s a different kind of beast.
Building something people rely on for their education is a heavy responsibility. It’s not like an app you can just delete if it’s buggy. People’s futures are, in a small way, tied to your uptime and your feature set. After scaling to 50,000 active users, I’ve seen what works, what breaks, and what is just plain wrong when it comes to personalized learning. These aren’t theoretical lessons from a textbook; they were learned through server crashes, angry user emails, and the occasional, wonderful thank-you note.
Here are the seven hardest lessons I learned.
1. Personalization Is Not Just a {user.name} Variable
In the early days, we thought personalization was just about greeting a student by name on their dashboard. We were proud of it. It felt personal. It was also completely useless. True personalization isn’t about surface-level tricks; it’s about fundamentally altering the learning path based on who the student is. It’s about knowing that one student struggles with abstract concepts but excels at practical application, while another is the exact opposite.
Our first real attempt at this was a disaster. We built a complex algorithm that sorted content based on a 20-question onboarding survey. It took us three months. On launch day, we found that over 80% of students just clicked random answers to get through the survey. The “personalized” paths were garbage.
The real breakthrough came when we stopped asking and started observing. We started tracking how long it took a student to complete a module, what resources they clicked on, and where they paused or re-watched a video. We used that behavioral data, not a questionnaire, to build a profile. When a student was flying through algebra but struggling with geometry, our system would automatically surface supplemental geometry exercises and dial back the algebra difficulty. That’s when the magic happened. Engagement shot up by 40% in a single quarter.
2. Engagement Is a Daily Battle, Not a One-Time Victory
You don’t just “achieve” engagement. You fight for it every single day. We had this one feature—a point system with badges—that we were sure would solve engagement forever. It worked for about two weeks. Students farmed the easy badges and then the novelty wore off. The engagement graph looked like a pump-and-dump crypto chart.
What we learned is that sustainable engagement comes from intrinsic motivation, not cheap gamification. It comes from a student feeling a sense of genuine progress. The most powerful feature we ever built was a simple “learning map.” It was a visual representation of the skills they had mastered and the ones that were up next. It wasn't a game, but it was compelling. It showed them how far they had come and gave them a clear view of the path ahead.
We also learned that social accountability is a powerful force. We created small, opt-in study groups of 4-5 students who were at a similar point in the curriculum. They could share notes and ask questions. We didn’t even build a chat feature at first; we just facilitated the groups. The simple act of knowing others were on the same journey was a huge motivator. These groups had a 30% higher completion rate for modules than students who studied alone.
3. AI Is a Scalpel, Not a Sledgehammer
When you’re a tech guy from Silicon Valley, you want to throw AI at every problem. We certainly did. We had grand visions of an AI tutor that could have Socratic dialogues with students. We spent a fortune on it. The result? A chatbot that mostly said, “That’s an interesting question. Can you rephrase it?” It was a glorified search engine with a personality, and students saw right through it.
Responsible and effective AI in education is about precision. It’s a scalpel, not a sledgehammer. Instead of trying to build a know-it-all oracle, we found success by applying AI to very specific, narrow problems.
For example, we used it to analyze student-written essays. The AI wouldn’t grade the essay, because that’s a job for a human who understands context and creativity. Instead, it would provide instant feedback on grammar, sentence structure, and repetition. It would highlight passive voice or suggest stronger verbs. It was a tool, not a teacher. It took care of the mechanical aspects of writing, freeing up the human teacher to focus on the student’s ideas and arguments. That’s a responsible use of AI. It augments the human, it doesn’t try to replace them.
4. The “Boring” Stuff Is the Foundation for Magic
No one gets excited about database optimization or API latency. As a product-focused founder, I certainly didn’t. I wanted to be in Figma, designing cool new features. But when you have 50,000 people hitting your platform, the boring stuff is everything. A 200-millisecond delay in page load can be the difference between a student staying for a 30-minute session or bouncing after 30 seconds.
I learned this the hard way. We launched a new interactive quiz feature. It was beautiful. But we hadn’t properly indexed the database table that stored the quiz results. When 5,000 students tried to take a quiz at the same time after a marketing push, the entire site ground to a halt. The query time went from 50ms to 30,000ms. We were down for four hours. It was the most stressful day of my life.
After that, we changed our entire philosophy. We dedicated one out of every four engineering sprints to “boring” stuff: performance, security, and reliability. We became obsessed with metrics like p99 latency. It didn’t produce flashy new features, but it built trust. The platform became fast and reliable. And that reliability is the invisible foundation upon which all the magical learning moments are built.
5. Listen to the Quiet Students
Your loudest users are not always your most insightful. For every student who sends a detailed bug report, there are a thousand who just get frustrated and leave without a word. Finding a way to hear from those quiet students is one of the hardest and most important things you can do.
We started doing something that doesn’t scale: I personally emailed 10 students every single day. Not a marketing email, but a real one from my own account. I’d say, “Hi, I’m Sahin, the founder. I saw you just finished the module on calculus. How was it? Was there anything that was confusing or that we could do better?”
Nine out of ten wouldn’t reply. But the one who did would give me pure gold. They’d say things like, “The video was great, but the accent of the instructor was hard for me to understand,” or “The third quiz question didn’t seem to match the material in the lesson.” This was feedback we would never get in a forum. It was raw and honest. We added subtitles to all our videos because of that feedback. We fixed dozens of small-but-critical content errors. You have to seek out the quiet voices.
6. Vanity Metrics Will Kill Your Mission
In the world of startups, we’re obsessed with growth. Daily active users, sign-ups, time on site. These are the numbers you put in your pitch deck. For a while, we were obsessed too. We would celebrate every time our user count ticked up. But then we looked closer and saw a disturbing trend: our user count was going up, but the average number of completed courses per user was going down.
We were getting a lot of people to sign up, but we weren’t actually helping them learn. They’d log in, poke around, and leave. We were a glorified content library, not a learning platform.
That realization was a turning point. We changed our primary success metric from “Daily Active Users” to “Weekly Learning Progress.” We defined this as a student completing at least one full module and scoring above 80% on the assessment. This metric was much harder to move. It forced us to stop focusing on marketing hacks and start focusing on the core product experience. It aligned our business goals with our students’ goals. Chasing sign-ups feels good, but chasing learning outcomes is what actually builds a meaningful company.
7. The Team Is the Product
This is the biggest lesson of all. You can have the best ideas, the most advanced tech, and a huge market. But if you don’t have the right team, you have nothing. In the early days, I hired for skill. I wanted the best engineers, the best designers. But I learned that in a mission-driven company, passion for the mission is more important than any specific skill.
Our best product manager wasn’t a product manager by trade; she was a former high school teacher. She didn’t know how to write a perfect spec doc, but she had a profound, intuitive understanding of how students think. She fought for the user with a passion I’ve never seen before. She was the soul of our product.
Building a team of people who genuinely care about education is the ultimate competitive advantage. They will work harder, solve problems more creatively, and stick with you through the tough times. The code we wrote five years ago is mostly gone, replaced by better technology. But the culture we built and the team that embodies it—that’s the thing that has lasted. The product is just a manifestation of the team.
Building a learning platform was a journey of a thousand mistakes. But these seven lessons are the ones that have stuck with me. They apply to any EdTech company, but they also apply to anyone trying to build something meaningful. It’s never about the fancy tech. It’s always about the people you’re trying to serve and the team you build to serve them.
Frequently Asked Questions
How were these items selected?
Each item on this list comes from direct experience, either from building my own companies or from patterns I've observed across the 200+ startups I've invested in. I prioritize practical, actionable items over theoretical concepts.
Are these recommendations still relevant in 2026?
Absolutely. While specific tools and tactics change, the underlying principles remain consistent. I update my thinking regularly based on what I'm seeing in the market and across my portfolio companies.
How do I know which items apply to my situation?
Start by honestly assessing where your biggest bottleneck is right now. The items that address that specific constraint will give you the highest return on your time and energy.