Bootstrapped founders know that every line item in your SaaS bill cuts into profit. The hosted analytics service you signed up for looked cheap at first, but now you’re paying per gigabyte ingested, per query, per custom metric. You’ve got months of customer data, but you hesitate before running a new analysis because you know it will cost money. Meanwhile, you still don’t have the dashboards you actually need.

The other approach is to build analytics from scratch. You recalculate numbers on every dashboard refresh, which is fine until you get actual traffic. Then every refresh becomes slow, and every slow dashboard is a feature you’re not shipping because you’re spinning waiting for a query.

The default assumption in web development is that serious applications run on serious databases, and serious databases means a separate server process, connection pooling, user management, and a configuration file that will eventually be wrong in a way that takes an afternoon to diagnose. MySQL and PostgreSQL are excellent databases. They are also, for the median small site, a solution in search of a problem — infrastructure designed for concurrency, scale, and replication requirements that don’t exist at any traffic level the site will realistically see for years.