How I think about headless CMS in 2026

Gautam Khorana · 2 May 2026 ·5 min read

The three questions I ask first

When a client asks me to pick a headless CMS, I do not start with feature lists. I ask three questions in order.

1. Who edits content, and how often?

If the answer is "one founder, a few times a month", the editor experience matters less than developer experience. Markdown in a Git repo might be fine.

If the answer is "a marketing team of 5, daily", the editor experience is the entire project. The CMS has to feel like Notion or it will not get used.

2. What is the content shape?

Simple blog posts with a title and body? Almost any CMS works.

Structured product catalogues with relationships, variants, pricing tiers, regional availability? Now we are in CMS territory where Directus and Sanity pull ahead of WordPress.

Heavy media-driven storytelling with custom block types? Sanity's Portable Text wins. Directus's M2A relations are second best.

3. Where does the content go?

If the answer is "one website", any CMS works.

If the answer is "website plus iOS app plus marketing emails plus partner API", you need an API-first CMS. WordPress can do it with extra work; Directus is built for it natively.

What I recommend most often

For 70% of mid-market clients, Directus + Astro or Next.js. Self-hosted on Railway or DigitalOcean, costs around $10 per month, gives editors a Notion-like experience, ships fast.

For high-design editorial work, Sanity + Next.js. Pay the per-document cost; the editor experience is worth it.

For headless WordPress, Faust.js or a custom Next.js consumer. Use this when the client refuses to migrate off WP.

What I avoid

Contentful Enterprise (price-to-value bad below 100k records). Strapi v4 (active churn). Anything with "AI-native" in the headline (not yet ready for production blog workloads in my testing).