30 days to build a micro-SaaS Product of the day on Product Hunt

As with most of my side-projects, this one started off as a solution to my own problem.

I built it, made it scalable because why not, added billing, and published it on ProductHunt where I had exactly 3 followers.

I never expected it to reach 10 upvotes, let alone 500. I could not have imagined it ever reaching Product of the Day, but that's exactly what happened.

The process

Day 1 - ideation

I had built this website with around 50 thousand pages around 3 years ago. I had always hoped that it would get indexed eventually by chance but that wasn't happening.

I went to the Google Search Console, and started requesting indexing for my pages. I thought I had cracked the problem, so the next logical step would have been to either build a Selenium script or go all-in with a headless browser automation.

As I was clicking on that "Request indexing" button happily and contemplating possible edge cases for my headless browser automation, a horrible popup appeared. I had reached the daily limit of 10 indexing requests.

Roadblock. Go through it or go around it.

So I looked for a solution and I found the official Google Indexing API.

I took the entire day to create an overview of the architecture needed for my new project based on Google's documentation; no fancy software, no Domain-Driven Development, just a ton of sketches on a few pieces of paper.

Day 2 - proof of concept

The core solution itself is really straight forward: assign a Service Account to your GSC, have it run every 24 hours with at most 200 URLs in a batch, then regularly check for updates.

Where things get kind of tricky, though, is building this thing in such a way that thousands of users communicate with Google's servers without any race conditions, networking bottlenecks, time or space complexity issues, security issues and it all needs to be fully transparent.

I chose PHP as the backend language because I've missed it. It's been a while since I last wrote something in PHP and I wanted to see how 8.x is treating me. I'm quite pleased with how the language has evolved over the years.

The proof of concept was an incredibly ugly command-line interface script that only did one thing. It sent batch requests to Google. And it worked!

No website, no SOLID principles, no test-driven development - nothing. Just a horribly beautiful working piece of software that didn't scale at all and had zero documentation. But what it did was prove there's a fix to my problem.

Day 3 - buy the domain name

It obviously didn't take the whole day to think about a domain name, but that's all I did on day 3.

I went on GoDaddy, typed a few queries, and picked one that looked kind of funny. The nice thing was the domain was on sale as well, so I could justify this purchase to myself more; because, you see, like many of you - I have bought way too many domains over the years.

So "add to cart", purchase, done. I spent the rest of the day trying not to thing about whether I'd let this domain expire as well a year from now.

Days 4 to 29 - it's go time!

A little bit of context is probably required here. I've been writing software for about 15 years now. I've built many things in many different languages, and "I have a particular set of skills" - insert Taken meme here -, and a few ready-built components.

I was very jealous of my time so I ordered myself to reuse building blocks from previous projects as much as possible.

Reused components

• For the front-end: reused a Material UI theme for Next.js
• For the back-end: reused an SDK generator I had built for OpenAPI v3 a couple of years ago
• Billing: translated a library I had written for LemonSqueezy from JavaScript to PHP
• Authentication: reused Firebase components

Shortcuts

• I did not use test-driven development, except for the billing component
• I did not use Laravel or Symphony because I felt they bloated my project
• I drew the logo in PhotoShop in 2 minutes - including the time to open and close the program; for past projects I would have used Illustrator for the 2D version and Blender for the 3D version with various transitions and animations.
• I wrote the Router component of the API from scratch because it took less time than integrating a ready-made solution; it helps if one of your hobbies is Regular Expressions.
• I did not waste any time on the design; just added the texts first, then "decorated" everything with what felt right at the time; I had spent weeks designing mockups in previous projects.
• I built the REST API methods just-in-time as I was using them in the hydrated statically generated frontend
• "No bells or whistles" - the software just needs to guide the user through these steps; that's it!

Fun

If you spend just 1 month building something, it doesn't get old yet. It still feels like fun, it still feels like play.

I had built many project before that spanned years and after a certain point you begin to feel like this thing is a black hole that will eat-up all of time and it will never be done. They say "all projects are always at most 80% done". Funny until it's not funny anymore.

Forcing yourself to keep the timeframe really short for a project is tough, but it guarantees the project will be fun.

So what do I mean by fun, here's an example: the cron task manager that spawns every minute decides which tasks to execute in that slot and then launches the appropriate threads (or processes, in some scenarios) to handle the workloads.

The fun thing here is that all tasks use something that I called "epoch" - borrowing the term from web3 -, which is nothing more than an integer between 0 and 1439, the "minute of the day".

The fun and interesting thing is that it's really difficult to really get the true minute of every single day because of: DST, time zones, leap years and random acts of God. So the solution here was to calculate the "epoch" starting from 0, using the Unix Timestamp as reference. That way there are no leaps, no gaps, and no race conditions. Isn't that nice?

I know - your definition of fun may be different, but I like the simplicity here.

The launch

I didn't publish on ProductHunt right away. I really didn't think it would make any difference since I don't have the following I've heard is necessary to have a successful launch.

After the project was done, I used it for my own websites. And it worked!

A few days had passed and I had convinced myself that nothing was truly, horribly wrong. So I published my thing on PH and went for a walk. I was notified that the launch would happen the next day at 10 AM my local time.

So the next day I refreshed the page expecting to see 0 upvotes. Shockingly though, 9 people had already voted for my tiny SaaS. One hour later there were 50 upvotes. Then 100.

The startup finished on the 3rd place, earning a badge for December 9th 2023.

Conclusion

Before this SaaS I had built a game engine from scratch in 2 years; I launched that on PH as well and got only 6 upvotes.

So let's recap: 2 years of work yielded 6 upvotes and 30 days yielded 500.
The conclusion here is that effort does not necessarily correlate to perceived value, I would say. What do you think?

I would love to hear your thoughts and, of course, any late votes still count - so click on that upvote button 😉.