I built a tool for my kid’s exams a year ago - it just crossed 3,032 tests with $0 marketing.
About a year ago, my kid asked me a question: "How can I practice for my PSLE (Singapore primary school) listening and oral exams online?" I looked around and realised there wasn't actually anything dedicated to those specific formats.
So I built a simple, frictionless site for her: pslepractice.com.
Twelve months later, the project has grown entirely through organic discovery. It just crossed 3,032 tests completed with 790 unique sessions. Top Google searches are now leading students directly to the site for queries like "psle listening comprehension practice" and "ai oral exam practice."
A few things I learned about "accidental" SEO:
-
Solving a gap is better than optimisation. I didn't do any traditional SEO - no backlinks or guest posts. The site ranks because it solves a specific problem that didn't have a solution. When Google sees people searching for things like "p6 listening comprehension practice online" and then staying on the site for 20 minutes to finish a test, it learns the site is the right answer. Intent-match beats keyword-stuffing.
-
Frictionless access builds trust. I made the choice to keep it free and sign-up-free. No paywalls, no gated results. For a student who needs practice, that lack of friction is everything. It turns a tool into a utility that people actually feel comfortable sharing.
-
Moving from answers to guidance. I recently added "Ollie," a Socratic AI tutor. Instead of just giving the answer when a student gets something wrong, Ollie guides them through the logic to help them find it themselves. It’s only been used about 15 times so far, but it's the direction I'm most excited about. AI shouldn't just do the work; it should help the student learn how to do it.
-
The value of "Passive" projects. I haven't "managed" this project in months, yet it continues to serve 15+ tests a day. It’s a good reminder that if you build something genuinely useful for one person, there are usually plenty of others with that same need.
I’m keeping this 100% free. I don't want to create a barrier for deserving students who just need the practice. It’s been a rewarding way to give back to the community and a great sandbox for learning how to build AI tools that people actually find useful.
I like that the product has evolved from helping students practice to helping them think.
The part that stood out most wasn't the traffic—it was adding Ollie to guide students instead of simply giving answers. That's a fundamentally different philosophy. Practice tools measure performance, but guidance tools build understanding, and that feels like a much stronger long-term direction.
that’s exactly what I was aiming for. It’s actually much harder to prompt an AI to be 'socratic' than it is to just have it give the answer, but the learning value is so much higher.
Interesting.
Your reply made me think about the consequence of choosing to teach through guidance instead of answers.
I don't think where that choice leads is obvious at first, and I don't think I can explain the reasoning properly in a thread without leaving out the parts that actually matter.
If you're open to it, what's the best email to reach you on?