Guide

Page builders vs custom themes: who really edits the site?

ELK Digital 10 min read

Part of the measured guide: WordPress speed optimisation — what actually moved the numbers.

Page builders like Elementor won the last decade for a good reason: a custom-built site cost more and took longer, so for most retail budgets the builder was the sensible choice. AI has changed that. A bespoke, hand-coded theme no longer costs three times as much or takes a month longer to build. But AI did not change the one thing page builders are good at: letting a non-technical owner edit their own site without a developer. So the decision was never about speed. It is about who edits and maintains the site after launch, and for how long. For a shop that is a serious commercial asset, rarely restructured by its owner, custom now wins. For an owner who rebuilds their own pages every week, a builder still earns its place.

That is the whole argument. The numbers below come from a rebuild we just completed for George Moss Jewellers, an independent jeweller in Uckfield whose site I have run for years.

Which should you choose?

The short version, before the detail:

A page builder still wins if the owner rebuilds their own page layouts often and wants to do it without a developer.
A custom theme wins if the site is a serious commercial asset, rarely restructured, and maintained by a developer who wants fewer plugins and fewer things that break.
Custom with a proper editing layer (Gutenberg blocks or fields) if you want the owner to edit content day to day without handing them full layout control.

The deciding factor is not speed. It is who edits and maintains the site after launch.

Why did page builders win in the first place?

Page builders won because, on a normal small-business budget, a custom build cost more and took longer. Choosing one was rational, not lazy.

A decade ago a bespoke theme meant weeks of a developer’s time and a bill to match. A page builder meant you launched in days off a template, and you could change a headline or swap an image without phoning anyone. Elementor, Divi and WPBakery built large businesses because they solved a real problem: most small firms could not justify custom development, and the builder gave them a professional site they could run themselves.

The strongest of those advantages is owner autonomy. Take it seriously, because it is the one thing the custom side has to earn back rather than assume.

What actually goes wrong with page builders like Elementor?

Page builders are not badly made. They are general-purpose, so they ship far more code than any single site needs, and that surplus compounds into slow pages, plugin sprawl and fragile updates.

A builder has to be able to build anything, so every page loads styles and scripts for features yours never uses, wrapped in nested container divs. Then, for the things the builder cannot do, you add plugins: a search-and-filter add-on, a reviews widget, a forms plugin, a snippets manager. Each one brings its own assets and update cycle, and each is one more thing that breaks when something else changes.

This is not theoretical. When we opened the database of George Moss’s WooCommerce shop, the page builder and its add-ons had left behind 50 stored template entries, more than 1,000 rows of builder metadata, 127 options and transients, and roughly 11 MB of generated CSS, before a single product photo had loaded. The shop ran Elementor, Elementor Pro and the Crocoblock Jet suite for its product search and filtering, alongside separate plugins for forms, reviews and code snippets. Several were doing jobs that a few lines of theme code now handle. Every one was a dependency that could fail on an update.

The builder bought to make the site easy to manage had become the main reason it was hard to maintain.

How has AI changed the build-versus-buy maths?

AI collapsed the one variable that justified page builders: the cost and time of a custom build. What used to take weeks of an expert’s time now takes a fraction of it. AI did not remove the need for the expert.

The old objection to custom was economic, not technical. Custom was cleaner and faster; it cost more to produce. AI-assisted development changes that arithmetic. An experienced developer with the right tooling can produce a bespoke theme in a fraction of the old time and budget. The reason most small businesses reached for a builder is the reason now disappearing.

Be honest about the limit, though. AI multiplies someone who already knows what good looks like. Point it at a build with no judgment behind it and it produces plausible, generic, subtly bloated output: the same problem as a page builder, in newer clothes. The value is not “AI built it.” The value is that an expert built it, faster, with AI doing the typing.

We tore Elementor out of a live shop: what it had been costing

When we rebuilt George Moss’s site off Elementor and Crocoblock onto a hand-coded theme, the homepage dropped from 134 requests to 45 and from 2 MB to 700 KB. It also got much faster. But speed was the side effect, not the point.

We did not do the rebuild to win a speed test. We did it to stop the fragility and the sprawl. The performance came along for the ride. Here is the same homepage, measured before and after on WebPageTest:

Measured on the homepage	Before (Elementor build)	After (hand-coded theme)
Server response (TTFB)	2.9 s	0.22 s
Largest Contentful Paint	3.9 s	1.4 s
Fully loaded	8.8 s	1.8 s
HTTP requests	134	45
Page weight	2.0 MB	0.7 MB

Largest Contentful Paint crossed from “poor” into Google’s “good” band, which is anything under 2.5 seconds. The custom filtering we built replaced the Jet plugins, so a whole category of dependency went away.

The part most case studies skip: the number that mattered most was not on the speed test. It was how little the site now needs from me. That is the real return on leaving a page builder. Not a faster score, but a site that stays built.

So who actually edits the site after launch?

This is the real decision, and the honest answer is usually the developer, not the owner. I have run George Moss Jewellers’ site for years. Checking it, updating it, keeping it working was never on them to remember. It was my job.

That changes where the cost of a page builder lands. Builders are sold on owner autonomy, and if an owner genuinely restructures their own pages every week, that autonomy is worth paying for. But when a developer maintains the site, the autonomy goes unused while the weight stays. Every plugin update is a risk I carry. Every odd layout break is a job on my desk. The sprawl is mine to track.

For the person maintaining it, a page builder is more surface area to go wrong. The rebuild was not about handing the client a slicker editor. It was about fewer moving parts, so there is less to break and less to maintain. Day-to-day content still edits in the standard WordPress editor; the structural work stays with me, as it always has. There is just far less of it now.

A bespoke theme is only as editable as you build it to be, so this cuts both ways: leave out a proper editing layer (Gutenberg blocks or structured fields) and you commit the owner to a developer for every change. That can be the right call or the wrong one. The point is to choose it deliberately, around who actually maintains the site, rather than defaulting to a builder for an autonomy no one ends up using.

Is a custom theme worth it in 2026?

Custom wins when your site is a serious commercial asset, rarely restructured by its owner, that needs to stay fast and stable for years. A page builder still wins when a non-technical owner needs to rebuild their own pages, often, without calling anyone.

A builder is the right call when you are launching fast on a tight budget, when the site is simple or short-lived, or when you will self-serve major layout changes week to week. Custom is the right call when the site earns money, when speed and stability matter, when you want fewer moving parts and a longer life, and when you are comfortable with content edits going through a maintained process rather than a drag-and-drop canvas.

The editability caveat resolves most cases. A good custom build can still hand you Gutenberg editing for your content. You give up redesigning a page from scratch yourself; you keep the control you use day to day. For most retailers, that is the trade they would choose if anyone laid it out plainly.

Should you let AI rebuild or redesign your website?

Yes, but only with a human expert directing it. AI is a force multiplier for someone who already knows what good looks like. On its own it produces work that looks right and is not: generic, dependency-heavy, and slow in ways you will not see until later.

The right question is not “AI or not”. It is who is holding the wheel. Tools change; judgment does not. A redesign is the moment to fix the foundations, not to swap one black box for another.

FAQ

Is Elementor bad for SEO or speed?

Elementor is not bad for SEO by itself, but its page weight and the plugins it leads to slow pages down, and slow pages hurt rankings through Core Web Vitals. The problem is rarely one setting; it is the accumulated weight.

Should I get off Elementor?

Only if it is costing you: slow pages, frequent breakage, a stack of plugins propping it up. If your site is fast and stable and you edit it yourself every week, stay where you are.

Can AI build a WordPress site?

An expert using AI can build a bespoke WordPress theme far faster than was possible a year ago. AI with no one directing it builds something that looks right and is not: generic, and bloated in ways you will not see at first.

Is a custom theme harder to edit than a page builder?

It can be, unless it is built with an editing layer. A good custom build gives you Gutenberg editing for your content; what you give up is drag-and-drop control over layout.

Will I be locked in to my developer?

Custom code is a codebase another developer can pick up and maintain. A page builder locks you into the builder and its plugins. Both are forms of lock-in; the custom one is the more portable.

What does a custom theme cost in 2026?

Less than it used to, because AI has removed much of the time that made it expensive. Not as cheap as a free template, but no longer the multiple-of-budget it was when the page builders won.

We measured the before-and-after figures above on the same homepage URL with WebPageTest, median of repeated runs. The “good” Core Web Vitals thresholds (LCP under 2.5s, INP under 200ms, CLS under 0.1) are Google’s, published on web.dev.

Related guides

If you’re weighing a rebuild, the honest place to start is a conversation — custom WordPress themes are the bulk of my work.

Thinking about a rebuild?

Let's talk about what your site actually needs.

Start a project