Jean Galea has made his living from WordPress for twenty years, and he just wrote that it is no longer his automatic first choice. He is right about the symptom and wrong about the framing. WordPress did not decline. It re-segmented. The default moved, the install base did not, and the work that stayed is the work worth having. That is a different story from the one the word "dying" tells, and it matters because the two stories lead to opposite business decisions.
Galea’s piece, “WordPress isn’t dying, it stopped being the default”, is worth reading in full because it is honest in a way most takes this year are not. He is not a defector throwing stones. He still builds on WordPress, he still picks it for the jobs it is best at, and he says plainly that he keeps reaching for other tools first. When the person with every reason to default to a platform stops defaulting to it, that is data. We run a WordPress agency, so we have the same incentive to read it the comfortable way, and we are not going to.
What Galea gets right
Steelman first, because the argument deserves it.
The on-ramp is gone. WordPress won the long tail because the five-minute install was the easiest way to get a site online. That is no longer true. You describe a site to an AI builder and it exists. No theme shopping, no plugin rabbit holes, no install. The single biggest reason WordPress became the default has been removed, and Galea is correct that the codebase makes this worse: twenty years of procedural PHP is hard for AI tooling to read and modify, so WordPress is structurally disadvantaged at the exact moment building moves to AI.
The original audience left. WordPress was built to democratize publishing. Two forces emptied that constituency. Google’s Helpful Content updates and AI Overviews cut the traffic that made an independent content site a business, so the blog-as-livelihood did not slow down, it switched off. And the people who just wanted a voice went to social platforms where there is no hosting, theme, or plugin to learn. The need that WordPress filled for millions of small publishers genuinely shrank.
The ecosystem consolidated. The owner-operators who used to fill WordCamp hallways, the plugin shops and small hosts, got rolled up into a handful of big companies. The enthusiasm you meet at events now is often an employee’s enthusiasm, which is real but different from a founder with everything riding on it. Galea is describing a roll-up, and roll-ups change who shows up.
And leadership earned a lot of the exodus. The clearest small example is meta trac ticket 6511, a request to restore the plugin active-install growth charts that WordPress.org stripped from the directory. Those charts are one of the only signals a plugin developer has to know whether their work is growing. The ticket is marked high priority. It was opened years ago, reopened, poked at, and left without a decision. That is the texture of the whole grievance: reasonable asks from the people who fill the directory, met with silence. The WP Engine saga was the visible part of something that had been grinding for years.
All of that is true. We are not going to pretend otherwise to protect a business model. The disagreement is not about the facts. It is about what they add up to.
The 40 percent is not inertia, it is a new center of gravity
Galea’s load-bearing claim is that market share is a lagging indicator. Installed base stays high out of inertia long after the story has changed, so the ~40 percent of the web running WordPress tells you about the past, while the flow of new projects tells you about the future. The new-project flow is moving away, therefore the share number is a corpse that has not cooled yet.
The first half is correct. Installed base does lag. The second half does not follow.
Inertia and switching cost are not the same thing. Inertia is “nobody bothered to move.” Switching cost is “moving is expensive and often irrational.” A brochure site nobody touches is inertia. A 4,000-product WooCommerce store with a custom checkout, a fulfilment integration, and three years of SEO equity is not sitting on WordPress because its owner is asleep. It is there because re-platforming would cost more than it returns and would put revenue at risk for months. Treating both as “lag” hides the distinction that actually predicts the future.
| Indicator | What it measures | What it says about WordPress |
|---|---|---|
| Total install base (~40% of the web) | Accumulated past decisions | Stable, slow to move, partly inertia |
| New long-tail projects | Blogs, brochure sites, first-timers | Leaving for AI builders, Astro, social |
| New control-critical projects | Enterprise, regulated, complex commerce | Staying, because the alternatives cannot do the job |
| Switching cost of existing complex sites | Cost to leave, not reason to stay | High and rising with site complexity |
Read that way, the share number is not one lagging blob. It is two different populations moving in opposite directions. The low-margin long tail is draining out, exactly as Galea says. The high-value control-critical base is sticky for structural reasons, not sentimental ones. A platform losing the first group and keeping the second is not declining. It is re-segmenting toward its defensible core.
From convenience to control
Galea makes the key observation himself and then walks past it. He notes that what is left, enterprise and WooCommerce, are “the cases where control beats convenience,” and calls it telling because convenience is what WordPress used to win on. That is the whole argument, and it points somewhere other than decline.
A platform that wins on convenience competes with every new convenient thing that ships, and right now that is a wave of AI builders. A platform that wins on control competes with a much shorter list: custom builds, and the handful of enterprise CMS vendors who charge for the privilege. The convenience market is brutal and getting more crowded. The control market is narrow, high-value, and hard to disrupt, because the thing customers want there, ownership of the code and freedom to extend without a vendor’s permission, is exactly what a hosted AI builder cannot offer.
WordPress losing the convenience layer to AI is real and probably permanent. But the convenience layer was always the low-margin commodity end. For an agency, the long-tail brochure site was never the good business. The re-segmentation pushes WordPress toward the work that pays: regulated content operations, publishers who need editorial workflow at scale, stores that need to extend behaviour a closed platform forbids. That is a smaller market and a better one.
When we still reach for WordPress, and when we do not
Galea lists his own choices: Astro for blogs and small sites, an AI build for heavy custom logic, Shopify for a store, WordPress for what it is best at. We make almost the same call, and the agency version of that list is the most useful thing we can hand a reader, because the default being gone does not mean the choice is hard. It means you have to make it on purpose.
| Project type | Default in 2016 | Deliberate choice in 2026 | Why |
|---|---|---|---|
| Personal blog, brochure site | WordPress | Astro or a static framework | No moving parts, fast, cheap, no maintenance surface |
| Just want a voice online | WordPress | Social platform | No site needed at all anymore |
| Heavy custom dashboard, app-like logic | WordPress + custom | AI-assisted custom build | Modern stack is easier for AI tooling to reason about |
| Standard online store | WooCommerce | Shopify | Hosted, predictable, lower operational burden |
| Store needing custom checkout, ERP integration, ownership | WooCommerce | WooCommerce | Extensibility and code ownership a closed platform blocks |
| Regulated or audit-critical content at scale | Enterprise CMS | WordPress | Own the code, satisfy policy, mature editorial workflow |
| Content operation with many editors and locales | WordPress | WordPress | Editorial workflow and multilingual tooling are still strong |
The pattern in the right column is the same one Galea identified: convenience work leaves, control work stays. Note that this site runs on Astro and Cloudflare, not WordPress, and we still run a WordPress agency. Those are not in tension. They are the re-segmentation, applied to our own stack.
What the fork talk actually signals
The strongest evidence against the “dying” reading is the thing Galea offers as evidence for the crisis: the fork proposals.
Joost de Valk, who built Yoast into a household name in the space, called for breaking the status quo and ending the project’s one-man rule. Ryan Hellyer wants to revive BackPress as a compatibility layer so WordPress can be rebuilt on a modern foundation, which is also the direct answer to the AI-readability problem. Malcolm Peralty laid out the case for splitting the project in two, a frozen WP Classic for the enormous existing base and a modern WP Next for everyone else.
Galea reads this as a symptom, and he is half right. It is a symptom of governance failure. It is not a symptom of decline. Nobody writes a serious modernization plan for software they have written off. You write “the case for the split” about something you love that you think is stuck. Dying platforms do not attract fork proposals from their most credible contributors. They attract silence and quiet migration. The fact that the loyalists are still arguing about how to fix the foundation is the clearest sign the foundation is worth fixing.
The risk here is not death. The risk is that governance stays frozen long enough that the modern-foundation work happens outside the project instead of inside it. That is a leadership problem with a known fix, and it is a very different problem from a platform that nobody wants.
The honest read from an agency that still ships WordPress
So here is the read from people with every reason to read it the other way. Galea diagnosed the symptom correctly and reached for the wrong word. WordPress did not decline. The web moved, the long-tail audience left, the on-ramp got eaten by AI, and the independents who built the ecosystem got bought or moved on. What remains is real, narrower, and more valuable per project than the commodity layer that left.
“Default” and “best choice” were the same thing for fifteen years, which is why losing the default feels like losing the platform. They have come apart. WordPress is no longer the thing you reach for without thinking. It is the thing you choose on purpose when control beats convenience, and that choice is still correct often enough to build a business on. We know, because we keep making it, and we keep being right about the projects where we make it.
The work now is not defending the default. The default is gone and arguing about it is a waste of a good decade of expertise. The work is being precise about which projects belong on WordPress and ruthless about the ones that do not, then doing the control-critical ones better than a generalist or an AI builder can. That is a smaller market than 2016. It is also a market where an agency that actually knows the platform is hard to replace, which is more than the convenience layer ever offered.
Related reading:
