I Asked Claude to Review the WordPress Site Editor as a Beginner

Environment tested: WordPress 7.0.2, Twenty Twenty-Five 1.5, fresh install. Scope: Appearance → Editor only — not wp-admin, not plugins, not themes.

Part 1 — Scorecard by surface

Beginner readiness out of 10. A 10 means a Canva-literate, non-technical person succeeds first time without help. A 0 means they cannot succeed at all without first learning WordPress.

Surface	Score	Verdict	Root cause
Landing screen	3 / 10	Five jargon words and one “Edit” button. No task, no starting point.	Interface design
Sidebar navigation / IA	2 / 10	Four of five items are WordPress concepts, not user goals.	Terminology
Pages	4 / 10	Best area, but exposes Slug, Author, Parent and “Discussion: Pings only”. No Home page exists.	Historical decisions
Templates	1 / 10	Named “Index”, “All Archives”, “Page No Title”, “Single Posts”; described using the template hierarchy.	Architectural complexity
Template Parts	1 / 10	Header and footer become locked purple boxes with “Edit original”.	Architectural complexity
Patterns	2 / 10	70 patterns, every one padlocked: “This pattern cannot be edited.”	Architectural complexity
Synced Patterns	1 / 10	The site’s H1 is a hidden synced pattern needing Edit pattern, then Exit pattern.	Architectural complexity
Global Styles	3 / 10	Well engineered, wrong audience. “Typesets”, eight unlabelled Aa tiles, Fira Code offered as a body font.	Interface design
Style Variations	3 / 10	Nine unnamed gradient swatches, no hover preview. Applying one silently forced the site title to uppercase.	Interface design
Block editing	6 / 10	Typing works well. Selection does not: two clicks needed once a template part is active.	Interface design
List View	5 / 10	Useful, but squeezes the canvas to ~40% width and lists “Paragraph, Quote, Paragraph, Quote”.	Interface design
Block toolbar	3 / 10	Floats over and hides the text directly above the block being edited.	Interface design
Inspector sidebar	3 / 10	Same block yields different panels in different contexts. Two panels both labelled “Navigation”.	Interface design
Saving	2 / 10	“1 site change… This change will affect your whole site”, a checkbox called “Title”, then Save again.	Architectural complexity
Publishing	3 / 10	No publish concept for design changes. Save and Publish mean different things in one product.	Hidden complexity
Header editing	1 / 10	Purple box → “Edit original” → focus mode → breadcrumb Template part › Header › Group › Row › Site Title.	Architectural complexity
Footer editing	1 / 10	Clicking a footer link selects the whole footer, then offers both “Edit original” and “Edit navigation”.	Architectural complexity
Navigation block	1 / 10	Editing the menu warns that “new pages will no longer be added automatically”.	Architectural complexity
Moving content	2 / 10	Dragging a block by its handle failed and instead selected an abstract “Content” wrapper.	Interface design
Responsive editing	1 / 10	Desktop / Tablet / Mobile is preview only. No per-breakpoint control at all.	Architectural constraint
Media insertion	4 / 10	Placeholder is fine; “Media Library” opens a 2012-era modal. Media tab offers only Openverse.	Historical decisions
AI features	0 / 10	None. Command palette search for “ai” returns fuzzy filename matches and links out to wp-admin.	Product gap
Zoom Out	2 / 10	Shrinks the canvas slightly, adds grey margins. No section affordances appeared. Purpose unknowable.	Interface design
State / URL / robustness	1 / 10	URLs are not addressable. A reset style change left a permanent “Review 1 change…” surviving reloads. Layout breaks below ~1300px.	Technical constraint
Weighted mean	2.4 / 10

Part 2 — The 20 biggest beginner pain points

#	Pain point	Severity
1	Header and footer are invisible when editing a page until you find “Show template”	Critical
2	“Edit original” — the original of what, and where does it live?	Critical
3	Saving asks you to approve a checkbox called “Title” under a warning about your whole site	Critical
4	All 70 patterns are padlocked: “This pattern cannot be edited.”	Critical
5	The site’s main heading is a hidden synced pattern needing Edit pattern, then Exit pattern	Critical
6	Templates are named after the WordPress template hierarchy, not anything a user recognises	Critical
7	A style variation silently changed the site title to uppercase, with no explanation	Critical
8	“Reset styles” left an unclearable dirty state and permanent “Leave site?” dialogs	Critical
9	There is no Home page; the homepage is a blog, and “Set as homepage” is buried in a row menu	Critical
10	Dragging a block to move it did not work; selection jumped to an abstract “Content” block	Major
11	Pressing Enter at the end of the page selected the Footer instead of adding a block	Major
12	Two clicks are required to select a block once a template part is selected	Major
13	Clicking a footer menu link selects the entire footer, not the link	Major
14	Responsive design is preview-only — nothing can be adjusted per device	Major
15	Editing the menu warns that pages will stop being added automatically	Major
16	The block toolbar covers the text immediately above the block you are editing	Major
17	The block library shows Math, Poetry, Preformatted and Pullquote before Image, and has no contact form	Major
18	Page settings expose Slug, Author, Parent and “Discussion: Pings only” to a florist	Moderate
19	Style variations and typesets are unnamed swatches with no hover preview	Moderate
20	The theme ships eight dead footer links: Blog, Events, About, Shop, FAQs, Patterns, Authors, Themes	Moderate

Part 3 — The five architectural decisions that create the most cognitive load

Decision	Why it costs so much	Load
Templates as first-class user objects	Forces the beginner to learn that a page is content poured into a separate reusable shell — and to learn the template hierarchy in order to read the names.	10 / 10
Template parts as separate records	Makes the two most obvious things on a website, the header and the footer, the two least editable. Creates “Edit original”, focus mode, purple boxes and permanent scope ambiguity.	10 / 10
Global Styles as a parallel, theme-scoped design system	Creates a second place where design lives, a second save flow, a second dirty state, and effects that appear on pages you never opened.	9 / 10
Patterns split into synced and unsynced	Two objects that look identical and behave oppositely — and one of them is invisible. Produces the padlock, the “Pattern” badge, and Edit / Exit pattern modes.	9 / 10
Everything is a block, including structure	Group, Row, Stack, Grid, Columns, Content, Query Loop and Template Part sit in the same tree as a paragraph, so selection and dragging become tree traversal.	8 / 10

Part 4 — The five changes that would most improve the beginner experience

Change	Impact	Effort
Ship a beginner mode that only ever shows Pages, Header, Footer, Menu and Site Style. Hide Templates, Template Parts, Patterns and the sync model entirely.	Transformative	High
Make headers, footers and menus editable in place, with a persistent inline notice (“this appears on every page”) instead of a purple box and “Edit original”.	Very high	Medium
Replace the save panel with plain-language confirmation — “Your site name will change everywhere. Save?” — and never surface a checkbox named after a database field.	Very high	Low
Rebuild the first ten minutes as a goal-led flow: what kind of site, pick a look with named previews, get a real Home page, then land in a page editor. Never land on an empty dark panel with an “Edit” button.	Very high	Medium
Fix the three trust-destroying mechanics: drag-to-move, single-click selection, and the unclearable dirty state.	High	Low

Part 5 — Strategic conclusion

Conclusion C. The architecture is fundamentally designed for advanced users and professionals. It cannot realistically become a best-in-class beginner experience without introducing a much simpler editing model on top of it.

Is the complexity mostly in the UI, or inherent in the architecture?

Inherent in the architecture, and this is the decisive finding. Roughly a third of what I catalogued is fixable UI work: the occluding toolbar, the broken drag, the unnamed swatches, the double-click selection, the dirty-state bug. But the worst moments are not UI failures at all. They are the interface honestly reporting what the data model is. “Edit original” exists because a header genuinely is a separate wp_template_part record. “This change will affect your whole site” exists because global styles genuinely are site-scoped. The padlock exists because an unsynced pattern genuinely has no editable instance. The breadcrumb Template part › Header › Group › Row › Site Title exists because that genuinely is the tree. You cannot design these away. You can only stop showing them — and the moment you stop showing them, the user loses the ability to predict what their next click will do.

Could Templates, Template Parts, Patterns and Global Styles simply be hidden?

Partially, and only at a cost. Patterns can be hidden almost completely; a beginner never needs the word, only “add a section”. Global Styles can be reduced to two controls, fonts and colours, with named presets. Templates and template parts cannot be hidden, because they are load-bearing on the primary task. Editing your logo, your menu and your footer is the job of building a business website, and in this architecture all three live outside the page you are editing. The concept leaks the instant the user tries to do the most obvious thing in the product.

Or do they inevitably leak into the editing experience?

Yes, and I watched it happen in a specific, repeatable way. Editing a page shows no header or footer at all until you discover a menu item called “Show template”. Reveal it and the header renders as an inert purple rectangle. Click into it and you are teleported into a different editing context with a different breadcrumb. Change one word and you are asked to approve a “site change” affecting your “whole site”. Four leaks, one task, thirty seconds — and every one of the five moments where a beginner no longer knows what is happening occurred inside that single interaction.

If WordPress were being designed from scratch today, would you build the editing experience around these concepts?

No. Not one of them. A 2026 greenfield beginner editor would have pages, sections, a site header, a site footer, a menu and one theme control, and it would treat templates, parts and sync as invisible implementation. These are not beginner concepts that were badly explained. They are professional concepts that were promoted into the primary interface because they were already there.

Where credit is due

This is an accomplished piece of software. The block model is coherent, Global Styles is a genuinely well-engineered design-token system, the pattern library is large and tasteful, and the theme output is clean and fast. Everything criticised here is a targeting failure, not a quality failure. For an agency building a client site, most of these complaints invert into strengths.

Why “radically redesign the interface” is not enough

I considered that conclusion seriously and rejected it on the evidence. A radical redesign of the current surface would fix pain points 10 through 20. It would not fix 1 through 9, because those are the model surfacing itself. To fix those you have to stop exposing the model — and that is not a redesign, it is a second, simpler editing product that treats the current Site Editor as its engine.

Confidence score

Dimension	Confidence
That the specific issues catalogued are real and reproducible	95%
That a genuine beginner would hit the top 10 within their first session	92%
That the complexity is architectural rather than cosmetic	90%
That Conclusion C is the correct call over Conclusion B	85%
Overall	88%

Deductions: one theme, one viewport range, one session — and an expert simulating a beginner rather than an observed cohort. Moderated testing with ten real first-time users would move this into the mid-nineties, and I would expect it to move the score down, not up. Experts consistently underestimate how early novices give up.

One-page conclusion

Could the WordPress Site Editor ever be the best beginner website editor in the industry, or would achieving that require a fundamentally different editing model?

It would require a fundamentally different editing model. Not different code. Different concepts.

The Site Editor is the most capable open-source design tool ever shipped inside a CMS, and it is aimed at the wrong person. Its entire vocabulary is inherited from how WordPress renders pages: templates, template parts, patterns, synced patterns, global styles, query loops, post content. A professional reads that vocabulary as power. A beginner reads it as a language they must learn before they are allowed to change the word “Blog” to “Bloom Florist”.

The test I applied throughout was the right one: could this concept disappear completely? For patterns, yes — call them sections. For global styles, mostly — two named controls. For the sync model, entirely. For templates and template parts, no. And that is precisely the problem, because those are the two concepts standing between a beginner and the very first thing they want to do, which is put their own name at the top of their own website. In a fresh install that task requires finding a purple rectangle, clicking a button called “Edit original”, navigating a five-level breadcrumb, and approving a checkbox called “Title” under a warning about their whole site. In Canva, Wix, Squarespace or Framer, it requires a double-click.

Canva, Wix and Squarespace do not win because their interfaces are prettier. They win because they made an architectural commitment WordPress has not: the object the user manipulates is the object they can see. Direct manipulation, immediate feedback, one place where things live. WordPress made the opposite commitment — separating content, structure and style into independently addressable records — and that commitment is correct for a platform and fatal for a first-time user.

So the honest answer is that no amount of interface work gets the Site Editor to best-in-class for beginners, because the interface is already telling the truth about the architecture, and the truth is too complicated. What would get there is a second editing model layered on top: pages, sections, header, footer, menu, one look. Templates still exist underneath. Patterns still exist underneath. The user simply never meets them, and is given a clearly marked door to the full editor when they outgrow the simple one.

WordPress has done this before. The 2003 answer to “how do I put words on the internet” beat every professional tool of its era because it was radically simpler than the alternatives, not more powerful. The Site Editor today is more powerful than its competitors and radically harder. Until a simpler model sits on top of it, the beginner who has used Canva and wants a florist website will keep choosing Squarespace — and they will be making the right decision.