Most of Northwind’s content is written by the person who knows the answer and published when they are finished. That is the right amount of process for a tasting note, and adding approval steps to it would be a way of making the site worse slowly.
One paragraph is different. The front page carries a note about what is on the roaster this month — what is landing, what has sold out, what is coming. It changes monthly, it is written in a hurry, it makes commitments about availability, and it is the most-read copy on the site. It is worth a second pair of eyes.
So: a review workflow, on that and nothing else.
Governance belongs to a place, not to a flag
The workflow is assigned to a folder — Seasonal — and everything in that folder inherits it.
Not a checkbox on the Component. That distinction matters more than it looks:
Governance is a property of where content lives, not a flag somebody remembers to set.
A per-item flag is a rule that is only enforced when somebody remembers to enforce it, which is to say the first month, and then a month when it is urgent. A folder is a decision made once about a category of content — put a thing there and it is governed, by construction, including the thing somebody creates next February in a hurry.
It is also how a reviewer knows where to look.
The consequence nobody expects
Here is the interesting part, and it is a modelling lesson rather than a workflow one.
The seasonal note was written inline on the front page — a paragraph inside the page’s own document, which is exactly what part 9 said such a thing should be. And it could not be reviewed, at all, until that changed.
You cannot review something that has no independent existence.
A workflow governs a Component or an Experience node. Inline content is neither: it has no version of its own, no place in the library, no separate publishing act. There is nothing for a review to be about — no artefact to move from Draft to In review, and no way to approve it without approving the entire front page along with it.
So the note became a Component, in the Seasonal folder.
And notice what that means for the inline-versus-reference question part 5 asked. The three tests then were: is it shared, is it separately publishable, is it worth finding in the library? The note is not shared and nobody will hunt for it — but "needs its own review" is a very direct form of "separately publishable".
Wanting something reviewed is a reason for it to be a Component. That is the same question arriving from the opposite direction, and it is the sort of thing that only shows up when you try to govern a real site rather than a diagram of one.
Choosing a shape, not inventing a state machine
A workflow has a kind, and the kind brings its vocabulary. For a publishing workflow, Draft and Approved are anchors — they cannot be renamed or removed, because the rest of the system knows what they mean. In between you add steps and name them; Northwind has one, called "In review".
That is a deliberate constraint and a kind one. A blank state-machine editor is a great way to build something that works for six months and then has two states nobody can explain. Choosing a shape and naming its middle covers what most teams actually need, and the parts that are load-bearing stay load-bearing.
A definition is small, and most of it is naming things you already have opinions about:
{
name: 'Seasonal copy review',
kind: 'publishing',
folder: 'Seasonal',
states: [
{ key: 'draft', templateKey: 'draft', name: 'Draft' },
{ key: 'review', templateKey: 'review', name: 'In review' },
{ key: 'approved', templateKey: 'approved', name: 'Approved' },
],
transitions: [
{ key: 'submit', name: 'Send for review', from: 'draft', to: 'review' },
{ key: 'approve', name: 'Approve', from: 'review', to: 'approved' },
{ key: 'send-back', name: 'Send back', from: 'review', to: 'draft' },
],
}
templateKey is doing the real work. Draft and Approved are anchors the kind supplies — they cannot be renamed away or deleted, which is what stops one team’s workflow from being unreadable to the next. Only the step in the middle is ours to name, and "In review" reads better on a badge than "Review".
Draft and Approved are the kind’s anchors and cannot be renamed away. Only the step in the middle was ours to name.
One practical note
Publishing completes the run. So the Component becomes editable again afterwards, and next month’s edit starts a fresh review rather than tripping over the last one.
That sounds obvious and is worth stating, because the alternative design — where a workflow run is a long-lived thing you have to remember to close — is a common one and it produces organizations where half the content is stuck in a state somebody left it in during an emergency.
Where this leaves us
One governed paragraph and a site otherwise unencumbered, which is about the right ratio.
Next: the Christmas range is decided in October and goes on sale in December, and somebody needs to see the whole site with it on before it does.
