Building for Humans, Recommended by AI: Lessons from 100 Silent Registrations.
It has been about one year since I started building SpellMe. I uploaded the first basic test version onto a web server around January 2025 and had a number of testers go through it in May. I launched the Beta 2 version in the middle of September, with a basic tweet, which was the first time members of the public were invited to use it. Fast forward to Jan 2026 and with very little marketing I have had thousands of visitors, hundreds of users, and of those, over a hundred have opted to register for an account.
Overall, I’m quite happy with the performance. I was not really intending to market last year, so wasn’t expecting people to find out about it. It has definitely been an interesting creation journey, though - one that I would like to share. I will be talking about how the current age of AI has changed the way people find things, like SpellMe and also about what I think is the most important part of any edtech app, which is research and real results.
Research
It was always important to me to ground SpellMe in real research. This research led me to write on topics, like the most important features for literacy apps to promote learning; types of fonts related to ease of reading and spelling; structured literacy and different ways of learning to read and spell; how user interface design can facilitate learning and help with accessibility; how AI can be beneficial; and the neurology of spelling and reading linked to neurodivergence. This research directly influenced the look and feel of SpellMe, and led to real results with users. This research will continue to show insights on how SpellMe can further improve.
Niche
From the very beginning, I have understood that a spelling practise app would be just for a small niche, which is why I was surprised that I started receiving any registrations as early as I did (even before my September launch). One reason to explain this is that even for a niche app, there is demand for help with spelling and literacy among parents and teachers. Over the last few months, that has become evident, however, the usual avenues of researching trends online in any sector have changed dramatically over the last couple of years. I can only relate this to how chatbots have changed people’s search behaviour. Before it was possible to track search words and link them to your site, through Google and other search engines. Although this is still possible, search on that avenue has dropped drastically.
The AI Chatbot Effect
According to a friend of mine in the enterprise/productivity software business, AI has had some quite significant effects. Their traffic from Google has dropped by over 80%, and the growth in traffic from Chatbots like ChatGPT and Gemini, is considerably less (around 20%). SpellMe has not been around long enough to feel the negative effects of this change, but I have definitely noticed that an overwhelming majority (over 70%) of my registered users are coming from ChatGPT. This means that most of my registered users find me through intentional searching, and SpellMe comes up in the AI chatbots as a solution to whatever their specific needs may be. This is promising. The problem is, unlike with people searching on Google or other more traditional search engines, there is no way for me to find out the exact problem these people have. I have to assume that the information my users get from chatGPT is accurate and it is not telling them that SpellMe is great for casting spells. Whatever the case is, the conversion rate which I will talk about next proves that I am doing something right and I will chalk it up to the research and development work that I have put into the app.
Building a habit
The lack of marketing that I mentioned before was intentional. Last year, I wanted to focus on bug fixing and feature improvements suggested by the users that were finding SpellMe, instead of getting more people to use the site that wasn’t up to my standard. Even so, my stats still showed that 1 in 6 users that did show up, by whatever means, chose to register. That’s a conversion rate of 16%, which is quite high for a niche edtech app. However, a high registration rate is only the first chapter of the story. As we move into 2026, I aim to shift my focus from simply welcoming new users to making sure they get value from coming back. After all, my research found that short regular practice sessions (about 10 minutes a day) were effective in improving spelling performance. I aim to turn that initial curiosity into a lasting educational habit, not for any monetary value but so users get the most out of what they signed up for.
Next steps
After some testing of the newest feature, the Teacher Dashboard, which I am really excited about, the app will finally come out of Beta release. Until then, I will be doing some more research on literacy, and will release the final Beta version (Beta 3) in the next few days. To be honest, it is available now, and has been for the last week or so, but in my typical marketing fashion, I have to send an underwhelming tweet to make it official. Look out for it and let me know your thoughts.