Most editorial pain isn’t about the CMS. It’s about the gap between how editors describe their work and how the schema is actually shaped. Update the launch banner sounds simple until it touches three documents, two singletons, and a feature flag — and suddenly publishing requires a meeting.
Our rewrite started with vocabulary, not code. We sat with the people who write copy, captured the words they actually used, and tried to make every one of them resolve to a single, obvious place in the studio. Where a concept needed to live in two documents, we made one the source of truth and the other a derived projection — never two co-equal copies.
The second shift was previews. Not just see the page, but see exactly what will be published, in the layout it will be published in, on the device it will be read on. When the preview matches reality, editors stop asking for engineering help on every change. Confidence is the real productivity unlock.
We also got serious about validation. A required field that explains why it’s required is dramatically better than one that just throws a red border. Schema messages are micro-documentation; written well, they replace a tab in Notion that no one reads anyway.
None of this is a finished story. As the team grows, the model will need to grow with it — more roles, more workflows, more guardrails. But the principle holds: editorial systems should bend toward the language of editors, not the other way around. Every time we’ve followed that compass, the ship lanes got shorter.
