Until 25 October, agenciarse.com was a textbook WordPress site: Elementor, the Wavo theme, Polylang for languages, Redux for options, and WooCommerce with WooCommerce Subscriptions (Stripe) to bill our maintenance plans. The public site is no longer WordPress, and the obvious question is whether that is "headless WordPress" and whether it is right for your business. The short answer is that ours is not fully headless, and that you may not need it either. If you want to see what we do with EmDash and Astro websites, that is the service page; what follows is what we learned on the way.
What is headless WordPress? Think of a corner shop
Picture a shop with a window and a back room. In a classic WordPress site both live together: what you edit and what the customer sees come from the same place, and every time someone visits, WordPress builds the page on the spot, with its theme and its plugins.
"Headless" means cutting the head off. You keep the back room (where content is written and stored) and the window is built separately, with other technology. The visitor no longer talks to WordPress. They get a site that was prepared in advance and is served as is, with nothing assembled on each visit.
That is the pitch: more speed, a smaller attack surface, total freedom of design. All of it is true. But the shop still needs someone who knows about window displays, and that is where the serious questions begin.
What you gain, no frills
Speed is the most visible gain. A page that is already built does not query a database or run PHP to appear. In our case it is the sum of several things: static HTML, served from Cloudflare's network, with very little JavaScript. The measurement from 14 November 2026 (PageSpeed Insights on mobile, Lighthouse 13.6) gives the home page a performance score of 98, with an LCP of 1.27 seconds, a CLS of 0 and a TBT of 45 milliseconds. The maintenance page also scores 98, with an LCP of 1.10 seconds.

Two things worth saying at once. First: this is lab data. We do not have field data yet, the data from real users. That comes from Chrome's CrUX report, which only publishes figures for a page or a domain once it has enough visits from Chrome users, and our site does not have them yet. Second: we have no reliable measurement of the previous WordPress site, so we will not tell you it "was twice as slow". We do not know. What we do know is that a well-optimised WordPress site can also score well. The difference is how much work it takes to get there and stay there.
You also gain security, in the sense that less is exposed. The public server no longer runs WordPress with its plugins; it serves static files. That is not immunity, it is less surface. And you gain freedom: you no longer depend on what a theme or a page builder allows, and the same content can feed a website and an app tomorrow.
What you lose, which is quite a lot
This is where brochures get shy.
Editing stops being "open Elementor and change it". On our service pages the text lives in the code, and every change is a commit that publishes itself. For us it pays off, because we work with code every day and the automatic deployment puts it live without manual steps. For a business with no technical team, one that wants to change a price or a photo on a Tuesday at nine in the morning without calling anyone, it can be a step back. The blog does have an editing panel, because we needed one there (EmDash, which we covered in another article).
Plugins on the public side disappear. Forms, sliders, pop-ups, a Yoast-style SEO plugin: on the static side there is nowhere to install them, and each one gets solved with code or an external service. If your site lives off front-end plugins, you are rewriting what you already had.
And you run two systems instead of one. More pieces, more places for something to fail, more things someone has to understand.
When it is worth it
A marketing site with a lot of traffic, or one where speed really weighs on sales, is the clearest case. Also when whoever maintains the site (you, your team or your provider) works with code and is not scared of a repository. When the sales pages change rarely: four times a year, not four times a week. When you want to send the same content out through several doors, a website and an app, say. And when you worry about having less of your public site exposed.
When it isn't
If you edit daily with no technical people around and rely on a visual builder, no. If you run a complex WooCommerce shop, no either: rebuilding the checkout outside WordPress is a lot of work, with payment gateways, taxes and plugins that break along the way. If the budget is tight, even less, because both systems have to be maintained. And if your site lives off front-end plugins, as we said, you will pay for it in hours.
To decide between yes and no, the question is not "which technology is better?" but "who is going to touch this next month?". If the answer is "me, with the mouse", classic WordPress is fine. And fine for good reasons: a dashboard everyone knows, thousands of plugins and a single system to maintain.
Three ways to set it up
Classic WordPress | Fully headless | Hybrid | |
|---|---|---|---|
What it is | Everything in WordPress | WordPress only stores content; the site is built separately | Static outside, transactional in WordPress |
Editing text | With the builder, no code | With its own panel or in code | Depends on the part: code or panel |
Site plugins | All of them | None on the public side | Only on the WordPress side |
Shop | Full WooCommerce | The checkout has to be rebuilt | WooCommerce stays in WordPress |
Maintenance | One system | Two systems, more code | Two systems |
Usually fits | Businesses that edit daily | Content or marketing sites with a technical team | Marketing sites with a shop or subscriptions |
Our case: hybrid, and why we did not go all the way
The public site (home and service pages) is static Astro, served from a Cloudflare Worker that also hosts our AI assistant and a small API of our own. The blog runs on EmDash, a CMS built on Astro, in another Worker with its own database (D1) and its own file storage (R2), at /blog/ on the same domain.
We did not rebuild the shop. WordPress with WooCommerce still lives at shop.agenciarse.com and only handles checkout, "my account" and renewals. We took out Elementor, Wavo, Polylang and Redux, and gave it a minimal theme of our own. When someone clicks a sign-up button on the site, the WooCommerce checkout opens in a modal window (an iframe).

Why not all the way? Because we bill our maintenance plans as subscriptions with WooCommerce Subscriptions, and that depends on the checkout, the customer account and the renewals running in PHP. Rebuilding it outside WordPress was expensive and risky, for subscriptions that were already collecting payments. We preferred to change the public side and leave alone what takes payments.
What links the two halves is small. Prices are not typed by hand on the site: on every build they are read from WooCommerce, and saving a product in WordPress triggers an automatic republish of the site. So WordPress stays the source of truth for what each thing costs, without the public site having to talk to it on every visit. Strictly speaking, that is the closest we get to "headless": WooCommerce as a data store, not as the shop window.
What it cost us
None of this was free, and what is most valuable in a project like this are the traps nobody warns you about.
When we changed the WordPress address to shop.agenciarse.com, WooCommerce Subscriptions noticed the URL was different and went into "staging mode", which switches renewals to manual. If you do not spot it, renewals stop being collected on their own and it can easily go unnoticed. You have to log in to wp-admin and confirm this is the real site.
Stripe webhooks do not follow 301 redirects. If the webhook still points at the old address, the redirect is not enough and the shop stops hearing about payments. We had to create the new webhook pointing at the subdomain.
Redirects took work too: every old URL needs a 301, so you lose neither external links nor what Google had already indexed.
And email. When you move DNS to Cloudflare, you must copy the MX, SPF, DKIM and DMARC records first. Otherwise the domain's email stops arriving or starts landing in spam, and you find out when a customer asks why you never replied.
We will not talk about the migration's effect on search rankings yet: with so few weeks behind us, any figure would be premature. We will cover it once there is enough Search Console data.
So, is it right for you?
It depends less on the technology than on your day to day. If your site changes little, speed matters to you and you have someone who can maintain it, the idea deserves a conversation. If it changes daily and you edit it yourself, or you have a complicated shop, the sensible thing is to stay on WordPress and look after it well. And if you are somewhere in between, maybe the answer is ours: move out what stands still and keep in WordPress what takes payments.
If you want to know which of the three fits your case, tell us about it on the EmDash and Astro web design page and we will tell you honestly, even if the answer is to change nothing. And if a shop is your worry, first look at what can be connected and what cannot in our WooCommerce integrations. WordPress web design is still what we do most.