An honest account of failures, bugs, and late fixes on the road to building around.md — from rough design to database security issues. What every mistake taught me.
# The Mistakes I Made Building around.md — and What They Taught Me
Deciding to build a product solo — no team, no safety net — sounds inspiring. The reality is far less glamorous: you code at night, you are your own designer, your own marketer, your own customer support. And you make the mistakes alone too — a lot of them, constantly.
I have already told the story of how around.md came to be. But this article is the other side of it: the specific mistakes I made, and what they taught me. Maybe it helps someone building their own thing right now.
The first version of the catalog looked decent. Cards, photos, neat colors. But I spent weeks polishing pixels before understanding the key thing: users don't care about perfect spacing between cards if they can't find a restaurant near their house in 10 seconds.
I rewrote the filter logic three times. At first I thought beauty sells. It turned out speed and ease of search sell. Design matters, but it is secondary to whether the site actually solves a real problem.
At some point I added working hours for venues — a normal feature, you pick opening and closing time. Everything worked fine, until someone typed "open 24/7" into the field instead of an exact time. The whole edit form broke — the field just showed nothing instead of a time.
A tiny detail, but trust in a product is built from exactly these tiny details. Users always find the one edge case you didn't think of. Since then I test forms by intentionally entering strange, non-standard values.
A catalog with empty, generic place descriptions is just a list of names. I thought photos and ratings were enough. Turns out they are not — people need text that explains why this specific place is worth visiting.
I wrote a script that generates descriptions via AI using real visitor reviews. The first version of the script was generous — it overwrote already-good descriptions on every run, wasting money regenerating something that was already done right. I had to rewrite the logic so the script only touches empty fields. A simple lesson: automation without checks is just automated mistakes, faster.
The most uncomfortable moment. When I finally sat down to review database access permissions — it turned out some settings were too permissive. Not a disaster, but a potential door better closed sooner rather than later.
Security is not something a user sees, which makes it easy to push to "later." But "later" can arrive as a much more expensive and stressful problem to fix than if it had been closed from the start.
One day I discovered that on phones, the browser's back button behaved oddly — the page fully reloaded instead of instantly returning. The cause: a single line of code blocking the browser's built-in cache. I didn't even know that thing existed until I ran into the problem.
It is a classic pattern: you build big features and skip the small technical details that actually define how a product feels to use.
Mistakes are not failure — they are a sign you are building something real, not just a theoretical plan in your head. Every one of these problems made around.md better once it was found and fixed.
Today around.md has 1,000+ places across Moldova, with filters that genuinely help you decide where to go, and content that doesn't feel like a template. Not because I got everything right the first time. Because I wasn't afraid to redo it.
If you are building your own product or business website right now and don't want to repeat someone else's painful lessons from scratch — that is exactly why I also started Dricomm. Experience is only worth something if you pass it on.
*Want to see the result of all these fixed mistakes? Take a look at the around.md catalog →*