Strapi

Last reviewed Aug 26, 2026 · 757 words

Two decisions on day one determine whether Strapi ages well for you: start on Postgres instead of the default SQLite, and treat major-version upgrades as scheduled projects rather than surprises. Get those right and the most popular open-source headless CMS — 73,017 stars, MIT-licensed, shipping since 2015 — is a genuinely pleasant thing to run. Get them wrong and you will meet Strapi's two documented cons at the worst possible time.

Start on Postgres even for the prototype

Strapi's quickstart happily creates an SQLite database, and that prototype has a way of becoming production. Migrating databases later is possible but it is exactly the kind of chore that never gets scheduled, so point Strapi at PostgreSQL in the first hour (MySQL and MariaDB are also supported). The runtime itself is Node and TypeScript, listens on port 1337 by default, and deploys fine as a Docker container, on Kubernetes, or as a plain Node process — the catalogue's 1 GB RAM floor is realistic for a small instance, though the admin panel build step wants more memory at build time than the running server does. One architectural detail that pays off permanently: content-type definitions live as code in your project, so your schema belongs in git and moves through dev, staging, and production like any other change, not by clicking the same buttons three times.

The admin panel is why editors will accept your weird API project

Plenty of tools give developers a content API; Strapi's durable advantage is that it also gives non-developers a place to work. The content-type builder lets you model articles, products, or whatever your domain needs, and the panel renders a clean editing UI for it automatically, with role-based access control to keep the intern out of the delete button. On the API side you get REST out of the box and GraphQL via an official plugin, so the same content feeds a Next.js site, a mobile app, and that one legacy consumer nobody talks about. The plugin marketplace fills the usual gaps — media providers, SEO fields, editor customizations — which is most of why teams pick Strapi over hand-rolling an admin on top of a bare database.

Know where the MIT license ends before you promise features

The core you self-host is genuinely MIT and genuinely complete for most projects. But Strapi is a venture-backed company, and some capabilities — historically the enterprise tier's SSO and audit-log territory, plus the managed Strapi Cloud offering — sit behind paid plans. None of this is a trap; it is the standard open-core deal, and the free core compares well against the proprietary platforms it replaces (the full switching math is in our Contentful alternatives guide). The failure mode is organizational: a team promises "SSO for the CMS" in a security review without checking which tier that lives in. Check the pricing page against your requirements list before you commit, not after.

Major upgrades are the real total cost of ownership

Minor updates are routine. Major versions are not: the v3-to-v4 transition was effectively a rewrite for many projects, and while more recent majors ship codemods and far better migration tooling, a Strapi major is still a budgeted engineering task — more so the more plugins and customizations you carry. Three habits keep this manageable. Pin your versions and upgrade deliberately rather than riding latest. Keep plugins few and mainstream, because abandoned community plugins are the number-one thing that strands instances on old majors. And snapshot the database plus the uploads directory before every upgrade, because a CMS is one of those services where "restore and try again next weekend" needs to be a real option.

What I'd do

For a content API behind a website or app in 2026, Strapi is my default pick from the CMS category: Docker, Postgres from the first commit, schema in git, plugins kept to a shortlist, and one calendar entry per year that says "Strapi major upgrade, allocate a week". Choose Directus instead when the truth you are serving is an existing SQL database rather than editorial content, and choose Ghost when the project is really a blog or newsletter wearing CMS clothing. But for the general case — developers who want an API, editors who want a friendly panel, nobody wanting a Contentful invoice — Strapi's decade of momentum is the safe bet, and the maintenance tax above is the honest price of it.