In 2023 we wrote that headless WordPress only made sense in a narrow set of cases. Two years on, the analysis still holds but with new shades. The scene has matured: tools are better, practices more established, and the community has settled.
What got better
More reliable tooling
WPGraphQL is now a common dependency in serious projects. Releases are more stable, types better, and extension plugins (ACF, Yoast, WooCommerce) actually work. On the Next.js side, @faustwp/next.js helpers and official templates from the WordPress core team have closed the "first week" gap.
Preview and revalidate
The historical pain point — seeing previews before publishing — is now solved cleanly with Next.js revalidateTag + WordPress webhooks. Edit a post, WP fires a webhook, the frontend invalidates just that post. No more 30-minute caches, no more full-site static rebuilds.
Exposed block themes and patterns
With WordPress 6.6 and beyond, block patterns can be exposed via API in a reasoned way. Not yet "render Gutenberg blocks in Next.js perfectly", but a real step forward for editorial architectures.
What did not change
- Two deploys: two systems, two monitors, two failure points. Without a team that can run both, it is a permanent cost.
- SEO to re-implement: meta tags, sitemaps, breadcrumbs, schema.org. Libraries exist but must be chosen and maintained.
- Editor learning curve: the gap between "press publish" and "appears online" confuses users used to the classic model.
The three cases where we say yes without hesitation
- Digital publishing with significant traffic and Core Web Vitals as KPIs: a static/streaming frontend offsets the complexity.
- Highly customised e-commerce: WooCommerce for orders and customers, custom frontend for dedicated UX. The pattern Shopify Hydrogen wants to own — but with WordPress you get it without lock-in.
- Multi-channel: content feeding site + mobile app + digital signage. One source, many outputs.
The two cases where we stay monolithic
- Brochure sites under 50 pageviews/day: classic WP is 10x simpler and fast enough.
- Sites run by non-technical editors without support: the discipline a headless workflow demands isn't there, and forcing it is not worth it.
The emerging pattern: WordPress + Next.js + edge cache
The stack we most often use today is: WordPress (with WPGraphQL and ACF Pro) as backend, Next.js 15 App Router as frontend, deployed on Vercel or Hetzner, edge cache with tag-based revalidation. No full builds — ISR/streaming throughout. For a mid-sized editorial site, pages update 1-2 seconds after WP save.
Is it more complex than monolithic WordPress? Yes. But if the complexity is justified by traffic and UX, today it is a solid road. In 2023 it was "promising". In 2025 it is simply an architectural option among others.