How much freedom should our CMS give marketers to build new pages?

Give marketers less freedom than they ask for and more than most studios ship. Then decide the exact amount per page type, before the build starts.

The timing matters more than the setting. If you decide after launch, you are not designing guardrails, you are removing something people already use. That conversation is always worse. Someone has published a campaign page with a purple heading and 12 pixels of space above it, and now the fix reads as a punishment.

Most teams get this wrong in one of two directions, and both directions end with a site that looks nothing like the design that was approved.

A redesigned site does not decay because the design was weak. It decays because a hundred small publishing decisions accumulate in a system that permits all of them.

Give a marketing team a free-form page builder with a colour picker and a spacing slider, and within about a year the new site contains the old site again. Not the old visual style, the old habits: the seven-section landing page, the heading that is a different size on every page, the card grid that is a card grid on three pages and a stacked list on the fourth because someone dragged it wrong.

The opposite failure is quieter and more expensive. Lock every template so only design can change a layout, and marketers go around you. They buy an external landing page builder for the webinar, because the webinar is on Thursday and design review is next week. Now you have unbranded pages on a subdomain nobody controls, outside your analytics, with a cookie banner that does not match. In our redesign projects, this is the more common of the two, and it is usually discovered by accident.

The rule worth remembering: freedom you do not grant in the CMS gets taken outside it.

There are four useful levels of CMS freedom. Assign one per page type, never one per site. A pricing page and a conference landing page have nothing in common except a domain name.

Each rung says what a marketer can change without asking a designer.

1. **Fixed template.** Words and images only. Section order, spacing and layout are set in code. Fits pricing, product, legal and anything you measure or A/B test. If a page's conversion rate appears in a board deck, it belongs here.
2. **Ordered blocks.** Reorder sections and swap them from an approved set. Fits most marketing pages. This is the default rung, and where the majority of pages should sit.
3. **Free blocks with token-only styling.** Any arrangement of any approved block, but colours, type sizes and spacing come only from the design tokens. No hex codes, no pixel values. Fits campaigns and event pages, where speed beats consistency but the brand still has to hold.
4. **Open canvas.** Everything is editable, including custom CSS. Fits almost nothing. Use it for short-lived experiments, and set an expiry date on the page when you create it. A page with an open canvas and no expiry date is how a site starts looking tired.

Work through page types, not pages. Four hundred URLs usually collapse into eight or nine types: homepage, pricing, product, blog post, customer story, campaign landing page, event page, legal, careers.

For each type, answer three questions in this order. Who publishes it? How often? What breaks if it is wrong?

Then assign the lowest rung that still lets that person ship without asking design. Lowest, not highest. Freedom is a cost you pay in review time and inconsistency later, so buy only what someone actually uses.

A worked case. A B2B software company had a marketing lead publishing roughly two pages a month: mostly customer stories, occasionally a webinar page. She asked for a drag-and-drop builder. The honest answer was that at two pages a month she would not learn one well enough to use it safely, and every page would be her first page in six weeks. Customer stories went to a fixed template with fields for logo, quote, three metrics and body copy. Webinars went to ordered blocks. Nothing else changed. The result: she publishes in about 20 minutes instead of half a day, and no page has needed design review since. The builder would have cost more and produced worse pages.

Two patterns show up in almost every CMS we audit after launch.

First, there is always one block nobody has used, and one block used for everything it was never designed for. The unused block is usually the clever one, a testimonial carousel or a stats band with an illustration slot. The overused one is usually the plainest: a heading and rich text. Marketers reach for the block they understand, then bend it, pasting a table into it, or a list of logos as inline images.

Second, spacing controls are the first thing to be abused, and colour pickers the second. If you ship a CMS with a free spacing field, expect values like 37 and 118 within a month. This is why token-only styling is a rung on its own. A dropdown offering small, medium and large is not a lesser version of a number field, it is a different tool. Sizes come from your spacing scale, so no one can invent a value that does not exist in the design system.

When someone asks for a custom layout, the request is almost never about layout. It is a missing block. Log every one of these in a single list with the page it was for, and after three months you will see the two or three blocks your site actually needs.

Six months after launch, count blocks used per page type. If a page type uses one block, it should be a fixed template. If it uses nine, someone is building layouts by hand and you owe them a better block.

If one person publishes twice a month, you do not need a page builder. You need three good templates and a written note on what each one is for. Refuse to pay for a bespoke drag-and-drop builder before you know your real publishing volume. Look at your current site’s history: how many new pages went live in the last twelve months, and who made them? If the answer is fourteen pages and one person, the builder is the most expensive part of your quote and the least used part of your site.

Most teams overbuy flexibility and underbuy documentation. Ten blocks with no guidance produce a messier site than five blocks with one sentence each explaining when to use them and when not to. That sentence is cheap. Write it while the block is being built, when someone still remembers the intent.

There is also a case for spending more, and it is narrow. If you run six campaigns a quarter with three people publishing, the third rung earns its cost quickly, and the guardrails are what stop six campaigns becoming six visual identities.

A field in the CMS is a design decision that someone else will make, hundreds of times, without you in the room. Review it the way you would review a component: what it is for, what it must not be used for, and what happens when it is fed the wrong content.

Do that, and the site you approved is still the site you have in two years. Skip it, and the drift starts on the first page published after launch.