What I Learned from Building (and Abandoning) InternetGuzeldir
InternetGuzeldir started with genuine excitement. I poured energy into building it exactly how I envisioned: a 900-line Python script reading an Excel spreadsheet to generate a curated link directory. Development was fun. Filling it with content was engaging. But over the past few months, my enthusiasm has evaporated. Here's why.
Google Hates Directory Sites
A website means nothing without visitors. Revenue doesn't matter much for a Turkish-language site anyway, but I wasn't even getting traffic. After digging through SEO forums, the verdict was clear: Google actively dislikes directory-style sites and flags them as low-quality content. I even got rejected from Google Adsense outright.
Frustrating? Sure. But I get it. Algorithms need substantial text to properly index content. A bare list of links doesn't cut it.
The fix: Add an articles section that links to individual entries in the directory. For example:
- "5 Apps That Help You Work in Noisy Places"
- Links to: Noises Online, Deftonic, Soundscape World
This gives Google the context it craves while adding genuine value for readers.
Excel Was a Terrible Database
I got swept up in the no-code hype when I started this. Everywhere I looked, people were building sites from spreadsheets. I even made the Eleman project under the same influence.
Excel works fine as a database—until you need to write longer content. Cramming an article into a single cell is torture. I couldn't add proper descriptions to entries. Data validation? Required fields? Forget it. There's a great breakdown of why this doesn't scale here.
The fix: Use a proper content management system. Lektor makes sense since I already use it for this blog and can troubleshoot Python issues easily. Static site generation still appeals to me—easy hosting, no backend headaches.
Category Trees Don't Work
Category hierarchies only work when something fits cleanly into one bucket. I learned this the hard way. Take OpenRA—it's a macOS game, a Windows game, and a desktop game. Listing it under one category feels arbitrary and broken.
I should've used tags from the start. This StackExchange answer drove the point home.
The fix: Switch to tags. If someone wants Turkish-language Windows strategy games, they can select all three tags and find what they need. itch.io's browse page handles this beautifully.
The downside? Static generation becomes impossible. You need dynamic page generation. Which probably means bringing back Django.
What I Should Have Done
Categories are out, tags are inRebuilding from scratch with DjangoWriting SEO-friendly articles that link to directory entriesActually making something people can find
I abandoned the project. The original vision was right, but the execution needed more work than I was willing to invest. Sometimes knowing when to move on is the real lesson.
08/2023