Moving articles from one CMS to another is, technically, a well-understood problem. Headlines, body text, images and dates can be mapped and moved by script. What cannot be moved by script alone is the meaning held in the taxonomy: the tags, subjects, sections and series that describe what each piece is about.
Over twenty years, that vocabulary grows in every direction. Tags are created for a single story, spelled three different ways, used for different things by different desks, and left behind when a section closes. The migration is the moment to sort it out, and it is editorial work.
Audit the vocabulary first
Export every tag and category with a count of how many articles use it and when it was last used. The shape is usually the same: a small number of well-used terms, and a very long tail of terms used once or twice. The tail is where most of the decisions are.
- Keep it as it is
- Merge it with another term that means the same thing
- Rename it to match current style
- Split it where it has been used for two different things
- Retire it, keeping the articles but dropping the term
Give the decisions to editors
Developers can suggest likely merges by spotting similar spellings and overlapping usage. But whether two tags mean the same thing is an editorial judgement. Set up a simple review process where editors can see each term, a sample of articles using it and the proposed action, and approve or change it.
The tags are the asset. Treat mapping them as editorial work, not a clean-up script.
Protect the URLs that depend on it
Tag and section pages often rank well in search and attract links. When terms are merged or renamed, their old pages need to redirect to the new ones. Keep a record of every change so the redirect map can be built from it automatically.
Plan for the new model, not the old one
A migration is a chance to design a better structure: perhaps separating subjects, people, places and organisations rather than keeping them in one flat list of tags. That structure makes related content, topic pages and archive products much easier to build. But it only works if the old archive is mapped onto it carefully.
- Agree the new taxonomy structure before mapping begins
- Map the most-used terms first, where the impact is greatest
- Run the migration repeatedly on test data and check tag pages as well as articles
- Give editors a way to keep the vocabulary tidy after launch
