Multi-tenant SaaS architecture: a practical guide for decision-makers

What is multi-tenant architecture?

It is a design in which a single instance of an application serves many customers (tenants), with each customer's data separated so no tenant can see another's. This is what lets SaaS platforms offer low monthly prices and ship continuous updates to everyone at once.

Three models of data separation

• Shared database with a tenant ID: lowest cost and easiest to upgrade; requires strict controls on every query.
• Schema per tenant: stronger separation with manageable overhead.
• Database per tenant: the highest isolation and the highest running cost.

The approach we recommend

For most SaaS platforms we recommend a shared database with PostgreSQL row-level security, so isolation is enforced by the database itself rather than by application code alone — with a fully dedicated environment available for customers who need one.

When do you need a dedicated deployment?

• When your sector's regulations require data residency in a specific country.
• When your workloads are large and irregular.
• When you need deep customisations that don't suit other tenants.

Questions to ask any SaaS vendor

1. How is my data separated from other customers'?
2. Where is data hosted, and can I export all of it?
3. How often are backups taken, and when was a restore last tested?

Next step

If you are evaluating a SaaS platform, ask for written answers to the three questions above before you sign.