Why I Left WordPress for Payload CMS + Astro

As of this week, archondigital.com no longer runs on WordPress. It now runs on Payload CMS for content and Astro for the front end, and the old WordPress install is retired from the public domain.

That's not a small thing for me. This blog started on WordPress as archon-digital.com, then lost the hyphen a few months in. I even left once, for Jekyll in 2014 at archon.digital. By June 2016 I was moving back to WordPress at archondigital.com. So yes, I've broken up with WordPress before. Meanwhile I built client sites on it for over 20 years, most recently on the Bricks, EtchWP and ACSS stack.

I've also bet early on the web's next tools before. When I launched www.gov.ph on 26 July 2010, the day of President Aquino's first State of the Nation Address, I believe it was the first Philippine national government agency site to run on WordPress. Most agencies were on Joomla back then. President.gov.ph later claimed that first in a press release, but its revamped site only opened on 16 August 2010, three weeks after gov.ph went live. I was also one of the first here to put sites behind Cloudflare, at a time when the alternative was Akamai, a paid service that was expensive.

This post explains why I left WordPress, how I picked what came next, and what the actual move looked like. If you're one of my WordPress clients, stay for the last section.

A love-hate relationship

I started making WordPress sites in 2004, a year after it first came out. My first paid WordPress side gig came in 2006, for a martial arts club. I owe WordPress a lot. It paid my bills, built my name, and carried projects I'm still proud of. When a client needed a site by Friday, WordPress was how I said yes. For a long stretch of my career, it kept me afloat.

That's what makes this hard to write. Love for a tool is mostly gratitude, and gratitude makes you forgive a lot. I forgave the plugin that broke on update, the theme that fought me, the late night patching a site that got hit. I told myself that was just the job.

But somewhere along the way the forgiving turned into dread. I started to hate opening the dashboard to a wall of update notices. I hated explaining to clients why their simple site needed another paid plugin, another licence renewal, another security scan. Toward the end I didn't just tire of WordPress. I abhorred what maintaining it had turned me into: a caretaker of other people's code instead of a builder.

Then the people running the project started fighting in public, and the last of my trust went with it.

So this feels less like switching tools and more like the end of a long marriage. A marriage where we built a life together, which I don't regret, but I'd stopped recognising who we'd become, and where staying was eating up on my soul and costing me more than leaving. Then I saw my lawyer, Mr Good News; He got me joint custody and legal separation; I'm so happy that I can't stop crying; I'm laughing through my tears; I'm laughing through my tears...

That's from Sting's "I'm So Happy I Can't Stop Crying", and it's been on loop since I took the plunge this week. Relief and grief in the same breath.

The governance reasons

WordPress the software is open source, but Automattic and its CEO Matt Mullenweg control WordPress.org, where plugins and updates come from. The last two years showed how much that matters. And so, the bigger push (in my decision) wasn't code. It was watching the platform's leadership become a risk I had to explain to clients.

I'm not taking sides in a lawsuit. My point is way narrower. When one person's dispute can cut off plugin updates for hosted sites, or take over a plugin my sites depend on, that's platform risk. I'd rather my clients' sites not ride on it.

The numbers are drifting too. Per W3Techs data cited here, WordPress's CMS share slipped from 63.2% (July 2023) to 58.8% (September 2026). Still dominant, but no longer the default it was.

Why Astro. It ships static HTML and only hydrates what needs to be interactive. Its content layer pulls from Payload at build time, so visitors never wait on the CMS. Pages are built from blocks: each Payload block has a matching Astro component with the same name and fields. That gives me a page builder without a page-builder plugin. The same caveat applies here: Cloudflare acquired Astro in January 2026. It stays MIT-licensed and deployable anywhere, and I run it on my own server. I could easily have put both sites on Cloudflare Workers, but I had room on a server I already run, and I wanted to stay hands-on with tuning and improving it. Given how early I bet on Cloudflare, I don't mind who owns it now.

How it runs. The CMS is on its own server. Each site is its own Astro repo on a second cheaper server behind Cloudflare. I could've easily used Cloudflare workers/pages but decided not to for this time. Shared blocks live in a private package both sites install. Hit publish in Payload and the site rebuilds, then Cloudflare clears just that site's cache.

The move itself happened over a weekend, but the planning took far longer, because a lot was at stake. In fact, this entire exercise was a dry run for a much larger public-sector project, where auditability and stability matter even more. More on that when it launches (real soon).