I wrote a dev tool post that got 25k reads. My follow-up got 11. Here's what I learned.

#testing #documentation #git #marketing

Back in 2018 I wrote a Medium article about a school project I built to help clean up GitHub repositories. I had published the little app and shared it with classmates. Little did I know that people were using the tool!

The way I found out was this: I published some other crappy article on Medium and wanted to see how many people had read it. It was something like 6. But I couldn't help but notice there were over 10k reads on the dev tool one. And a bunch of comments about how it's broken and it sucks. Woo hoo.

So I immediately fixed all the bugs and responded to the comments.
After that I polished up the app, made a new UI, and generally got obsessed with the the project again.

Funnily enough, even after that wake up call I didn't think about starting a db to track usage. I eventually did in 2022, but I lost 4 years of it's most popular time period. The db went down in 2024 and I didn't notice for like 9 months. Jeez. There were a lot of hard lessons learned here.

Eventually I got all the basics covered. But at that point the need for the tool died down significantly. I've tracked that we've had 6.5k unique signups for the RepoSweeper between 2022-2024 + 2025-2026. In reality the number is prob north of 15k.

Not many people squander lightning in a bottle as regularly as I do.

I recently expanded the tool suite to do other bulk actions that I saw in Github's Community Discussion board: collaborator management, visibility setting, archives, etc.

Anyway. I did a writeup about the new features and documented everything I'd built.

11 reads.

Not 11k. Eleven.

The painful irony is that the product is genuinely more useful now. But "more useful" doesn't make for a better headline.

The original post worked because it was about the reader's problem, not my product. "25 ways my tool helps you" is always worse than "here's the exact shell command I was too lazy to remember."

Every good dev post probably has one job: make the reader feel seen before you make them feel sold to.

I built something that solved a real problem for me, wrote honestly about it, and 25k people related. Then I got excited about what I built next, wrote about that, and almost nobody cared — because I switched from their perspective to mine.

Going back to basics. Next post: one problem, one solution, one story.
Hopefully I'll notice when it blows up this time.
RepoSweeper is still free if you want to check it out. RepoRecap PRO is the AI layer I got too excited about. Roast me in the comments.

Roast me in the comments.