We spent 2 months building a microservices architecture for a product with 20 users. I regret it.

When we started building our SaaS, I wanted to make sure we did everything "right." I read all the tech blogs, looked at how big companies like Netflix and Uber built their stacks, and convinced my co-founder that we needed a highly scalable, event-driven microservices architecture hosted on Kubernetes.

It took us 8 weeks just to get the infrastructure running. We spent weeks configuring Docker containers, setting up API gateways, and writing Terraform scripts. I felt like a genius. We were ready to handle millions of users on day one.

Fast forward six months. We currently have 20 active users, and our infrastructure costs are bleeding us dry. We are paying AWS $450/month for a product that generates $150/month in revenue. Even worse, every time we want to push a simple update (like changing the text on the landing page), it takes three days of wrestling with CI/CD pipelines.

Last week, I finally snapped. I spun up a monolithic app on a single $12/month server, migrated the core logic in two days, and we are shutting down the Kubernetes cluster. The new setup is blazing fast, and I can deploy updates in 30 seconds.

I fell into the classic trap: optimizing for scale before we even had product-market fit. I built a rocket ship when we were still trying to figure out how to build a bicycle.

For other founders: How did you decide when your tech stack was "good enough" to launch?

Did you start with a boring monolith? And if you over-engineered early on, how long did it take you to untangle the mess and simplify?