Skip to content

Addresses and SEO

This page covers how a site’s pages are addressed and how search engines see them: content addresses and their patterns, redirects, breadcrumbs, the title and description of each page, the schema.org markup, the sitemap, robots.txt, the site’s RSS feed, canonical links, caching behind a CDN, and the counters and the cookie banner.

Most of it is under Configuration → Search and SEO in the admin: SEO (/admin/config/seo), URL aliases (/admin/config/url-aliases), Redirects (/admin/config/redirects) and Counters and codes (/admin/config/integrations). Breadcrumbs are under Structure → Breadcrumbs (/admin/config/breadcrumbs).

Absolute links - canonical links, og:url, the sitemap, the RSS feed, robots.txt, the structured data, the links in e-mails - use the site’s address. Set it once:

  • in the admin: Site information (/admin/config/system/site-information), field Site address, for example https://www.example.com;
  • or in the config: public_base_url / PUBLIC_BASE_URL (the configuration). It overrides the admin’s field (which then shows the config’s value, greyed out), so a copy of the production database on a staging site keeps the staging address.

The value is an origin only: http(s):// and a host, no path. With neither set, the sitemap, the feed and robots.txt take the address from the request, canonical links and og:url are bare paths, the pages carry no structured data, and the account e-mails aren’t sent (see Email).

Every content item has an address (its slug), for example article/my-first-post; the page is at https://www.example.com/article/my-first-post. Addresses are built automatically from the content type’s URL pattern.

Set the pattern on the content type’s settings (field URL pattern) or for all types at once on URL aliases → Patterns (/admin/config/url-aliases/patterns). An empty pattern means [node:type]/[node:title]. Under each type the page lists the tokens its pattern can take.

Token Value
[node:type] the type’s machine name
[node:title] the title, transliterated to Latin letters
[node:created:Y], [node:created:m], [node:created:d] the creation date’s year, month, day (the site’s time zone)
[node:field:<machine_name>] the term in a taxonomy field of the type: the term’s own address without its vocabulary’s prefix (so with its parent terms)

Only taxonomy fields can be used in a pattern; text, number and other fields can’t. The pattern’s own text may contain lowercase Latin letters, digits and -, _, ., /. The SEO templates’ syntax works here too: [a|b] takes the first token with a value, {…} is a part dropped when a token inside it is empty (see Syntax below).

Examples:

Pattern Address
empty article/my-first-post
news/[node:created:Y]/[node:title] news/2026/my-first-post
[node:field:section]/[node:title] guides/basics/my-first-post (the term “Basics” under “Guides”)

A token with no value leaves no empty segment. An address that is taken gets -1, -2… at the end. Cyrillic letters are written in Latin ones (щ - sch, ё - yo), as in uploaded files’ names, and accents are dropped (é - e); a title that leaves nothing (no Latin or Cyrillic letters, no digits) gives the type’s machine name and a timestamp.

The patterns page saves all types at once: if one pattern is wrong, nothing is saved and the error shows under that type.

Content of a module’s own type (a forum topic, a poll) gets its address from the module; such types have no pattern.

In the content form, the Address section (shown to users with the administer content permission) has:

  • Generate the URL alias automatically - on: the address is made from the pattern when the item is created, then it stays - renaming the item or changing its terms doesn’t move it. Off: the address is the one typed in the field below; ticking it again on an address typed by hand makes the address from the pattern at once.
  • Slug (URL) - the address without the leading /. On a new item it may be left empty to generate it. A typed address another page of the site already answers - a View’s page, a term’s page, one of the engine’s own (login) or a module’s - isn’t taken: the content would hide that page, or never be reached.

Users without administer content don’t have the section, and an address, meta tags or comment settings they send anyway are ignored.

New content follows the pattern unless an address is typed. Whenever a saved item’s address changes, by hand or by the pattern, the old address redirects (301) to the new one, so links to it keep working.

Every item is also reached at /node/<ID>, whatever its address: that link redirects (302, as the address may change) to the item’s current address, for whoever may see the item. The action log links to content this way. An item whose own address is node/<number> keeps it.

A changed pattern applies to new content. Existing addresses stay as they are until you rebuild them.

URL aliases → Addresses (/admin/config/url-aliases) lists every content address with its title and type, and whether it is set by the pattern or by hand. Filter by a part of the address (Address contains) or by type; the pencil (Change) opens the content’s edit form.

Terms have their own addresses, set by their vocabulary: see Taxonomy.

After changing a pattern, apply it to the content that already exists:

  1. Open URL aliases → Rebuild (/admin/config/url-aliases/rebuild).
  2. Choose the content type. To make hand-set addresses follow the pattern too, tick Also the addresses set by hand (they will follow the pattern).
  3. Click Show what changes. Nothing is written yet: the page shows how many addresses would change and the first changes, Now and Will be.
  4. Click Rebuild the addresses. On a large site the work stops when the request’s time runs out; click Continue to carry on from there.

Every changed address redirects (301) from the old one. The content’s “updated” time doesn’t change.

Redirects (/admin/config/redirects) holds the site’s permanent (301) redirects: a request to From is sent to To. The engine adds one whenever a content address changes; add your own for moved or removed pages.

  • Add redirect: From - a path on the site without the leading / (articles/old-slug); To - where to send it: a path on the site with the leading / (/articles/new-slug), or a full address. Both are saved as typed.
  • Bulk import: one redirect per line, either from -> to or an Apache line RewriteRule from$ to [R=301,L]. Empty lines and lines starting with # are skipped. Here a leading / is removed from From and added to To if missing.

The list below shows the redirects by From, 50 to a page; the search finds a part of either address.

A redirect matches the path only: the query string isn’t compared, nor passed on.

Which old addresses need a redirect? The 404 log module adds a Not found (404) tab to this page: the addresses visitors asked for and the site didn’t find, the most asked for first, each with a button that adds its redirect here.

Breadcrumbs are built from the site’s menus: there’s no separate tree of sections, the menus are the tree. The trail always starts with Home.

  • A page whose address is a menu link - a View’s page, a term page, the search page - gets the trail of that link and its parents. It needs no rule.
  • A content item gets its place from the first matching breadcrumb rule: the rule’s menu link and its parents, then the item’s title. This holds even when the item’s own address is in a menu; without a matching rule it gets Home › title.
  • Anything else gets Home › title.

When an address is in several menus, the deepest trail wins; at equal depth, the menu with more links. Neighbouring links with the same address collapse into one, and disabled menu links (with everything under them) are skipped.

For example, with Guides › Basics in the main menu, /guides/basics linked from Basics and a rule “Article with the term Basics in Section → Basics”:

Page Breadcrumbs
/guides/basics (a View page in the menu) Home › Guides › Basics
/tags/go (a term page not in a menu) Home › Go
an article in the section Basics Home › Guides › Basics › the article’s title

Open Structure → Breadcrumbs (/admin/config/breadcrumbs). A rule has:

Setting
Content type required
Term field optional: a taxonomy field of the type. Empty (any content of the type) - the rule matches all content of the type
Terms with a field: the rule matches content that has one of these terms in that field, or a child term of one
Menu link any enabled link of any menu; the content’s trail is this link and its parents, then the title

To add a rule, choose the type, optionally the field, and the menu link, then click Add. If you chose a field, the rule’s page opens to pick the terms (Ctrl/⌘ to pick several); a rule with a field and no terms matches nothing. After changing the field, save the rule to refresh the list of terms.

The first matching rule from the top applies: drag the rows into order and click Save order. Put narrow rules (with terms) above the general rule for the same type.

A rule whose menu link was disabled is marked link not found or disabled and doesn’t apply. Deleting a menu link or a field deletes the rules that use it.

Pages with breadcrumbs also get them as schema.org BreadcrumbList structured data in the page’s <head>, when the site’s address is set.

The <title> and the meta description of content pages and term pages, and the og:image of content pages, are built from templates with tokens. Everything is on SEO (/admin/config/seo).

Note that the title is used as is: the engine adds no site name to it. To have one, put [site:name] in the template, for example [node:title] | [site:name].

For a content page, the title and the description are chosen separately. The first one with a value wins:

  1. The content item’s own value - Page title and Description (meta description) in the Meta tags section of its edit form. Leave them empty unless the item needs a hand-written title: a copy of the generated one would stop following the templates.
  2. The first matching rule by terms (see below).
  3. The content type’s template - Content types table, the pencil (Change).
  4. The default template - Default for all content.
  5. The built-in value: the title - the content’s title; the description - the beginning of the text.

If the chosen template renders empty (it has nothing but tokens, all of them empty), the built-in value is used. The description is stripped of HTML and of the text’s code blocks, and cut to 160 characters.

Image (og:image) is a template of its own, on the default and on each type (not on rules), usually an image field: [node:field:cover_image]. It gives the original image’s address, made absolute. Empty - the page has no og:image.

A rule gives its own title and description to content of a type with certain terms, for example “Article in the section Reviews”. It matches the same way as a breadcrumb rule: a content type, optionally a term field and its terms (child terms match too).

  1. Under Rules by terms, choose the content type and, optionally, the term field; fill in Title and/or Description; click Add.
  2. If you chose a field, the rule’s page opens: pick the terms and save.
  3. Drag the rules into order and click Save order: the first matching rule from the top wins.

An empty template in a rule falls back to the type’s template.

Token Value
[node:title] the content’s title
[node:field:<machine_name>] a field’s value: text as is; terms - their names separated by commas; a date - day.month.year (15.05.2027); a number as is; an image - the original file’s path
[node:field:<machine_name>:year] the year of a date field
[node:field:<machine_name>:iso] a date field in ISO 8601 (2027-05-15T10:00:00+03:00), for the schema.org markup
[node:field:<machine_name>:<format>] a date field in a format from Date formats (/admin/config/date-formats) by its machine name, for example [node:field:event_date:medium], in the page’s language
[node:summary] the beginning of the content’s text
[site:name] the site’s name

An unknown token renders empty. The template pages list the type’s fields with their machine names.

Syntax Meaning Example
[a|b] the first token with a value [node:field:local_date:year|node:field:date:year]
{ … } an optional part, dropped entirely if any token inside it is empty [node:title]{ ([node:field:event_date:year])} gives “Summer Fair (2027)”, or just “Summer Fair” without a date
{A|B} the first variant whose tokens all have values {[node:field:title_original] ([node:title])|[node:title]}

Inside […], | still means “or the next token”.

Under Term pages: a title and a description for the pages of all vocabularies (Title (all vocabularies), Description (all vocabularies)), and per vocabulary (the pencil, Change, in the table). A vocabulary’s template wins over the shared one; with neither, the title is the term’s Page title, or its name.

Token Value
[term:title] the term’s Page title (a field in the term’s form), otherwise its name
[term:name] the term’s name
[term:description] the term’s description
[site:name] the site’s name

The syntax above works here too. The term’s Page title also becomes the term page’s heading, so a term can keep a short name for labels and have a longer heading on its page.

Where Template Result
default [node:title] | [site:name] My first post | Example Site
type Event [node:title]{ - [node:field:event_date:medium]} Summer Fair - 15 May 2027
rule: Article, Section = Reviews [node:title] - review New Phone - review
vocabulary Tags [term:title] | [site:name] Go | Example Site

Canonical links, og:url and noindex aren’t templated; the engine sets them:

  • A content page’s canonical link and og:url are its address on the site’s address.
  • A list on page 2 and on is its own canonical page (?page=2), its title gets “- page 2”. Other non-empty query parameters stay in the canonical link.
  • noindex, follow goes on pages with no content of their own: the sign-in page and the two-factor pages, a user’s files, the search results, a View page with an active exposed filter; modules mark their own such pages. The other account pages are kept from crawlers by robots.txt.

The front page’s title is the site’s name and slogan, its description the slogan (both from Site information).

Search engines read a page better, and may show it as a rich result, when it says what it is in schema.org’s vocabulary (JSON-LD in the page’s head). With the site’s address set, breadcrumbs carry it, and so do content pages whose type is chosen on SEO - the Schema.org type of Default for all content or of a content type (As for all content, - none -, or its own):

Type For Its own properties
Article, News article, Blog post articles, news, posts Keywords, Section
Web page other pages Keywords
Event events Start, End (a date field with :iso), Place, Place’s address
Product or service goods, services, offers Brand, SKU, Price, Currency, Availability

The rest goes in by itself:

  • every type: the content’s title as its name (an article’s headline), the description and image the SEO templates above give the page, its address;
  • the articles and the web page: when the item was published and last changed;
  • the articles: the author (their profile) and the site as the publisher; an event: the site as its organizer.

A property’s value is a template with the content tokens, for example [node:field:tags] for the keywords or [node:field:start:iso] for an event’s start; an empty one is left out. A content type’s SEO page shows the markup of its newest item, with a link to Google’s Rich Results Test.

Site information has Who runs the site - for search engines: an organization, a local business or a person, its name (the site’s by default), its profiles elsewhere (an address per line), a phone and an email, a local business’s address; the logo uploaded on Site information is its logo (a theme’s logo isn’t). It is the publisher of the articles’ markup, and the front page carries it with the site itself. The site’s search in the results adds the site’s search (/search?q=…) for search engines to offer.

/sitemap.xml is a sitemap index, built on request:

  • /sitemap/static.xml - the pages of enabled Views (except those with a % segment in the address), pages that enabled modules add, and the links of the menus marked Include in the sitemap (a checkbox in the menu’s settings, see Menus);
  • /sitemap/nodes/1, /sitemap/nodes/2… - the content, up to 45,000 addresses per file, with their last-updated date.

Content is listed when a guest may see it: published, not in moderation, not in a closed section. Left out too: content of disabled modules’ types, of types with a Redirect to the address from field (see Content types), and the items used as the site’s error pages.

There’s nothing to set up besides the site’s address. Submit https://www.example.com/sitemap.xml to search engines; robots.txt names it too.

/robots.txt is built by the engine: for every crawler it disallows the admin, the sign-in and account pages, the content forms and actions, the search results, and the pages the enabled modules list; then it names the sitemap.

To add your own lines, open Configuration → Security → Access blocking (/admin/config/robots), section robots.txt. The text is added after the engine’s rules as written, for example:

User-agent: ExampleBot
Disallow: /

The same page blocks crawlers by User-Agent and by IP address. That blocks requests outright, unlike robots.txt, which crawlers may ignore.

/rss.xml is the site’s RSS 2.0 feed: the first 20 items of the View chosen as the Front page feed in Site information, the same list the front page shows a guest. Each item has the title, the link, the beginning of the text, the date and, if the View shows an image field, the image. The channel’s title is the site’s name, its description the slogan.

With no front page feed chosen, the feed has no items.

For more feeds - per section, in other formats, with their own field mapping - enable the Feeds module; it’s set up at /admin/config/feeds.

When a CDN or a caching proxy is in front of the site, it can keep the pages guests see. Set page_cache_max_age / PAGE_CACHE_MAX_AGE in the configuration to the number of seconds a page may be kept, for example 300.

With it set:

  • an HTML page answered with 200 to a guest (no session cookie, a GET or HEAD request) that sets no cookie gets Cache-Control: public, max-age=0, s-maxage=300: the CDN keeps it for 300 seconds, browsers ask again each time;
  • every other HTML answer - to a signed-in user, an error page, a page that sets a cookie - gets private, no-store and is never shared;
  • answers that set their own Cache-Control (the sitemap, the feed, the static files) keep it.

0, the default, turns it off. Guests see a change to the content only after the CDN’s copy expires, so keep the time short or purge the CDN’s cache after an important change.

Configuration → Counters and codes (/admin/config/integrations) puts external services on every page:

  • Site ownership verification - name=value lines, each a <meta name="…" content="…"> (Yandex.Webmaster, Google Search Console, Bing).
  • Yandex.Metrica - the counter number and its options (click map, external links, accurate bounce rate, session replay).
  • Google Analytics - the stream ID, G-….
  • Custom code - HTML/JS at the end of <head> and before </body> (an ad loader, a chat), only for the role with full access; the latter can stay off service pages (the admin, 404).

None of the counters and codes is on a page whose address carries a secret - a password reset link, a preview, an unsubscribe link (a token parameter or a {token} in the route): a counter would send the address on to its service. The verification metas stay.

The site’s own cookies - the session of a signed-in user, the protection of forms - need no consent. Analytics and ads do in the EU, and the banner asks for it. Visitors’ consent (cookie) on the same page:

Setting
Cookie banner on or off (off by default)
Mode Consent first (the default) - the codes of a group run only once the visitor allows the group; Notice only - they run at once, the banner tells about them and offers Got it
Banner text empty - the default text in the site’s language
“More” page the privacy policy: a path on the site (/privacy) or an address; empty - no link
Ask everyone again forgets every visitor’s choice - after the policy changed

Each code has its Consent group: Analytics (the default for Yandex.Metrica and Google Analytics), Advertising, or No consent needed (the default for the custom code). With Consent first the banner - a bar at the bottom of the page - offers the groups the site’s codes are in: Accept all, Necessary only, or Customize - a tick per group and Save the choice. Until then a group’s codes don’t run at all: they wait in the page and the browser runs them after the visitor’s yes. If no code waits for consent, there is no banner.

The choice is kept in the visitor’s browser for a year (the kosatka_consent cookie), so pages cached for guests stay the same for everyone. A link to #cookie-settings - say, a footer menu link with that address - opens the banner again to change the choice; taking a group back reloads the page.