An MVP analytics system that saves €8,000+ per month — without clouds or paid servers

Recently, a client came to us with a fairly typical situation for a European SaaS product:

They had an analytics system for ~1,000 users. And they were spending around €2,000 per month on hosting.

Turns out, their setup included around 10 cloud instances — for database, backups, logging, and services. It doesn’t sound like much in enterprise terms, but for an MVP, that kind of spend can burn fast.

We offered a radically different path.

Instead of yet another round of optimization, we rewrote the architecture from scratch:

• Switched to a queue + worker model
• Completely abandoned the cloud
• Deployed the system directly on office desktop machines

Currently, just 2 machines are connected — and that’s already enough to reliably process data for 5–6K clients, thanks to asynchronous architecture and scheduled load distribution.

Want more? Scaling to 4 machines could easily support a lightweight SaaS with 15–20K users, depending on the domain. (Think of something like a fitness tracker — regular background sync, but no real-time heavy lifting.)

This is what an MVP looks like when it knows how to grow without burning budgets.