CMS Architecture
Content structured so the people who write it can publish without a developer, and so the structure still makes sense in three years.
Why Payload?
Payload CMS is TypeScript, it runs inside the Next.js application rather than beside it, and the schema is code. Collections are defined in files that live in the repository, get reviewed like anything else, and version with the site. There is no separate hosted service holding the content model hostage, and no plugin marketplace quietly deciding what is possible.
What does content modelling actually involve?
Deciding what things are, before deciding what they look like. A service, a location, a project, an author, each with its own fields and its own relationships. Done well, a new location page is a row of data. Done badly, it is a copy-paste of an existing page and a slow divergence between the two.
How does a migration work?
Audit what exists, decide what survives, then map old structure to new. Most migrations are mostly deletion: sites accumulate pages nobody reads and taxonomies nobody uses. Redirects and site moves are mapped page by page, never wholesale to the homepage, because a redirect to somewhere irrelevant is read as a soft 404 and the accrued signal is thrown away rather than passed on.
Who ends up editing it?
Whoever should. Access is scoped by collection, so a copywriter edits copy and cannot restructure navigation. Fields carry help text written for the person using them, not for the developer who built them.