Templates and publishing: one template per category, one deliberate publish

The Content page is where a category gets its metadata. One template covers every page in the category; variables in braces fill in the parts that differ per page. Nothing you write here reaches your live site until you publish it, and publishing is its own tab.

The page in two tabs

Content is the editor. Categories are listed in the navigator on the right, the configuration for the selected category sits on the left, and the preview resolves that configuration against the category's real pages. Above the tabs are Undo and Redo, which step through your unsaved edits in the browser and nothing else.

The Content page with the blog category: the configuration on the left, the per-URL preview in the middle and the category navigator on the right.

Publish is the gate. It lists every page with content, says what would change on each, and asks you to confirm.

There is no Save button. Edits to a template are saved on their own about two seconds after you stop typing, and switching category saves the one you were on first. "Saving…" in the header is the only sign.

Pick a category

Categories are created on the URLs page, not here; with none defined the page sends you there. The navigator shows each category with a URL count and one of three states: Template means it has a template, Has content means its pages carry saved values without a template (a CSV import, for example), and Needs setup means neither.

The navigator: Whole site with 112 URLs, blog with the state Template, docs with the state Needs setup.

Whole site at the top of the list is not a fallback template. It is where the structured-data blocks that apply to every URL live; see Structured data.

Write the template

The Metadata section has three groups. Search snippet is the meta title, meta description and keywords. Social sharing is the Open Graph and Twitter title, description and image. Indexing is the canonical URL and the robots directive.

A variable is a name in braces, {title} or {h1}, and resolves to that page's own value. {name|fallback} uses the fallback where the page has no value, and {if:name}text{/if} includes the text only where the page has one. A name the system does not recognise is left exactly as written, and a page with an unresolved name in any field cannot be published; the preview marks those pages before you get that far. Variables explains where the names come from.

The Content section templates the H1, the H2s and image alt text. Those change what visitors see on the page, not just what search engines read, which is why the section opens with a warning.

Read the counter

Under the title and description fields the counter shows the real range across the category's pages once the template contains a variable: "49–98 / 50–60" means the shortest resolved title is 49 characters and the longest 98, against a good range of 50 to 60. A line under the field then counts the pages outside the range. A template without variables is simply measured as written.

The Meta Title field with the template, the counter 49–98 / 50–60 in red, and the line 50 of 64 pages over 60 characters.

The good ranges are 50–60 for the title, 120–160 for the description, 55–65 and 150–200 for the Open Graph title and description, and 55–70 and 125–200 for the Twitter ones. Below is short, above is likely to be cut off in results, and Google rewrites descriptions freely, so treat them as guidance.

A title and a description inside their ranges: 56 / 50–60 and 133 / 120–160 in green, each with a check mark beside the label.

A social field you leave empty publishes no tag at all. The page's own tag, if it has one, stays, and the platforms fall back on their own rules: Facebook to the page title, Twitter to the Open Graph tags. We never copy the meta title into an Open Graph title for you.

Robots for the whole category

Robots is a choice, not a template. Picking a directive writes it to every URL that is in the category right now. After that each URL owns its value: moving a page to another category does not change it, and a page that joins the category later inherits nothing. The select shows a value only when every page in the category agrees.

The Indexing group: the Canonical URL field and the Robots select reading Pick a directive, with the note Written straight to each URL.

Five directives are accepted: index, follow, noindex, follow, index, nofollow, noindex, nofollow and none. Follow the site (leave unchanged) means we publish no robots tag and the site's own stays in place, and that too is a choice we store.

A directive you pick here is confirmed at publish time: the confirmation dialog counts how many pages would go out with noindex, because a bulk noindex is the one content change that can cost a site all its traffic.

The preview

Each row is one page. Expanding it shows every field with where its value comes from. Live is what is published today, Pending is what the next publish would change, and a field marked as being removed means the page's own tag is used again. Not crawled and Meta only say how much we know about the page; headings and alt text come from the page body, which a Meta only page has not had read yet.

The first fifty pages are shown; Load more adds fifty at a time.

When a page keeps its own value

A page can carry a value of its own in any field: an inline edit from the preview, a CSV import, or the robots stamp all write one. A page's own value always wins over the template. The template is inert on that page in that field, and the editor says so under the field: "The template is not used on 12 pages with their own values."

To hand the field back, choose Use the template on all under the field, or Use the template on one page's row. The pages' own values in that field are removed and the template's text is used instead. What is already published changes only when you publish again.

Inline edits are seeded with the crawled value and the template's output, but only the fields you actually change are saved, so opening the editor on a page does not quietly make every field its own.

Publish

The Publish tab lists every page with content and its state: Live when what is published matches the draft, Changes pending when it does not, Not published when it never was, and Can't publish when something blocks it.

The Publish tab: the filters All, With issues and Ready, and rows with the state Live, one of them with the warning Title is 65 chars.

Two things block a page: an unresolved variable in any field, and a robots directive outside the five above. Everything else is a warning: a title under 10 or over 60 characters, a description under 50 or over 160, an H1 over 70, an empty title or description. Warnings never stop a publish, and you can mute a warning type from the Issues popover; muting is presentational and stays on this device.

Filter to Ready or With issues, select the pages, and click Publish. The confirmation dialog repeats the count, adds how many pages carry noindex and how many still have warnings, and then updates your live site. Pages that failed stay selected so you can retry them. Caches are purged and warmed on success, but it can take a few minutes before a visitor sees the change.

Unpublishing is not done here. The Deploy view can withdraw published content for the whole site or for a list of URLs.

After publishing

Open Deploy to see whether your integration is asking for what you published, URL by URL. Publishing and serving explains what Active, Idle and Error mean there.

The score does not move when you publish. It moves when the next crawl reads the page, so publish, crawl, then read the score.