How to keep a site editable without a page-builder mess

The honest version of custom website vs page builder is a trade about what an editor is allowed to touch. A company site stays coherent when people edit structured fields: the headline, the paragraph, the image, the offer. Absolute freedom on every pixel is a different product. It lets a Tuesday afternoon invent a layout the brand never had.

Both tools can publish words. They do not leave you in the same place a year later. Choose for the page you will still recognize, and for the person who has to change a sentence without a designer in the room.

What should be a field

List the things that actually change. On a young company site the list is short.

• The headline and the sentence under it.
• The proof: project names, a short description, a link.
• The people, if you publish them, with a photograph and a role.
• Posts or notes, with a title, a date, a summary, and a body.
• The primary action: the label and where it goes.
• Images that are content, with alt text stored beside the file.

Each of those is a field with a type. A headline is one line, not a nested row of font controls. An image is a file plus a description, and the crop is the one the layout already decided. A post is a document with a few marks for emphasis, headings, and links, not a board of draggable widgets.

Structured fields make the next edit safe. The editor fills in the offer. The page puts the offer in the place the design reserved. They cannot drag the headline into the footer, set it in a third typeface, or insert a countdown timer because a plugin made it look urgent. The urgency was the plugin's idea. The page's idea was the sentence.

Hanabi builds this way. The marketing site is a set of designed sections in production code, Next.js and Tailwind among the tools, and the words that must change sit in fields. The client portal holds brand files, documents, and updates after delivery. It is not a second, freer homepage. Editing the site and finding the logo are different jobs, and both stay calmer when they are not the same screen.

What should stay designed

The layout, the type scale, the color, the spacing, and the navigation are decisions. They were made so the first screen answers what the company does and how to talk to someone. A page builder that exposes those decisions as sliders will get new decisions every time someone is in a hurry.

Leave these in the design system and the code.

• Which sections exist, and in what order, on the homepage.
• The type pair, the sizes, and the measure of a paragraph.
• The color roles: background, text, a single accent, a quiet border.
• How a project card behaves on a phone.
• The menu, the footer, and the one primary action.

A new kind of page can still be added as a template, with its own fields, when the site has a real second job. A campaign with a different shape is a template. Unlocking every homepage section "just this once" is how the system becomes a scrapbook.

Google's guidance on helpful content asks whether a page looks sloppy or hastily produced, and whether a reader would leave having understood enough to act. A stack of mismatched blocks fails that reading even when each block was easy to drop in. People-first, in this case, means the editor's convenience does not outrank the visitor's ability to scan. Restraint in the editor is a kindness to the person who has never seen the brand guidelines.

Where a builder is the right tool

A page builder earns its place when the pages are disposable, numerous, and owned by people who will not have an engineer. A weekend of experiments, a temporary event page, an internal draft the public will never see. The cost of a messy layout is low because the page is going to be thrown away, and the cost of waiting for a custom template is high because the moment will pass.

Builders also fit a team that has accepted the trade in writing. They know the site will accumulate one-off sections. They have someone who cleans it up, or they are willing to live with the drift. That is a legitimate choice for a club newsletter or a personal project. It is a costly choice for the only page a buyer sees before they decide to write to you.

Judge the result with a loading measure, not with a slogan about which tool is modern. web.dev's good threshold for Largest Contentful Paint is within 2.5 seconds of the page starting to load. Extra scripts, late widgets, and images dropped in at display size all push that moment back. A custom page can fail the same test if it ships a heavy hero. The tool does not pass or fail. The page you published does. Open it on a phone and see when the offer appears.

If a builder is what you have, you can still behave as if you had fields. Agree on three section types and refuse the rest. Keep one type scale. Do not install a new visual plugin for a single page. Most of the mess is optional. The software permitted it. The team can decline.

Where a custom site is the right tool

A custom site is the right tool when the page is the company's public face and it has to stay recognizable. The design is specific. The sections are few. The editors change words and images inside a layout that was argued over and then finished.

The cost is real. A brand new layout takes a designer and an engineer, because it is a new template, not a drag. That cost is the price of coherence. You pay it when the structure changes, and you avoid paying a tax of confusion on every ordinary Tuesday.

Custom does not mean "only a developer can fix a typo." That is a failure of fields, not a virtue of code. The marketer should be able to change the headline, swap the portrait, and draft a post. The engineer should be the person who adds a section type. Choosing a CMS when the marketing team is small helps you decide whether those edits happen in a CMS or, for a site that rarely changes, in the repository with a preview. Either can be structured. Neither has to be a free canvas.

Ownership sits underneath the choice. The templates and the content should be yours. A builder subscription that holds the only copy of the homepage is a dependency with a monthly bill. A repository you can clone, plus a CMS you can export, is a site you can keep.

A practical way to decide this month

Write down the edits from the last quarter, or the edits you expect in the next one. If they are all words and images inside sections you already like, build fields and lock the layout. If they are genuinely different page shapes every week, and those pages are allowed to be temporary, a builder may be the cheaper mess.

Then write the refusals. No new typeface from the editor. No new accent color. No section that can be reordered on the homepage without a conversation. No plugin whose main feature is animation. Put those refusals where the editor can see them, in a short note at the top of the CMS or in the README.

Review one published page after a month of real edits. If it still looks like the design you approved, the fields are doing their job. If it looks like a bulletin board, the freedoms are too wide. Remove a freedom before you add a guideline document nobody will open.

Hanabi's services are the custom side of this trade: a designed site, production code, and an editing path for the things that change. If you are comparing a builder demo with a studio build, compare them on the edit you will actually make. Change the headline. Then try to ruin the page. The better system makes the first task easy and the second task difficult.