vidimus

seo: search engine basics

seo reads every built page and the site-wide crawler files and reports what keeps a page out of search results or makes it show up wrong: a missing title, a noindex left over from staging, a canonical pointing at another domain, a sitemap full of stale URLs. It is not in the default set.

Needs

Nothing to install. It reads the build (dist/) and starts no server or browser. Set the top-level siteUrl: without it the canonical, hreflang and sitemap checks can only verify that URLs are absolute, not that they point at your site and at pages that exist.

With render.mode on, every check reads the DOM a browser renders, so titles, descriptions and canonical links set by the app count.

Run it

npx vidimus seo

What it checks

Pages matched by the top-level exclude or by seo.exclude are skipped; by default that is the 404 page (/404.html or /404/). Pages with a <meta http-equiv="refresh"> are treated as redirect stubs and skipped too. The same problem on several pages is one finding listing the pages.

Every page

  • Language: <html> has a non-empty lang attribute. Missing: missing <html lang> (error).
  • Title: a <title> exists (missing <title>, error) and is within seo.titleLength (title outside 10-60 characters, warning, with the length of each page as detail).
  • Description: <meta name="description"> exists and is not blank (missing meta description, error) and is within seo.descriptionLength (meta description outside 50-160 characters, warning).
  • noindex: a <meta name="robots"> or <meta name="googlebot"> whose content contains noindex fails with noindex in the production build, unless the page path matches seo.allowNoindex. A noindex page still gets the language, title and description checks, but none of the checks below, and it is not counted as indexable.

Indexable pages

  • Canonical (seo.canonical): no <link rel="canonical"> is a warning (no canonical link); more than one is an error. The href must be absolute (canonical URL is not absolute). With siteUrl set, it must also be on that origin (canonical points outside https://example.com) and resolve to a page in the build (canonical is not a built page), with the base path of siteUrl taken into account.
  • hreflang: each <link rel="alternate" hreflang="…"> must have a unique language per page (duplicate hreflang "en") and an absolute href (hreflang URL is not absolute). Targets on siteUrl must be built pages (hreflang target is not a built page) and must link back (hreflang not reciprocated, with from → to as detail). Targets on other origins are not followed. x-default is treated like any other language value. This check has no switch.
  • Headings (seo.h1): no <h1> and more than one <h1> are warnings.
  • Duplicates: two or more indexable pages with the same title or description get duplicate title "…" or duplicate meta description "…" (warning). Pages that are all hreflang alternates of each other are exempt, so translations that keep an untranslated brand title do not trigger it.
  • Canonicalized copies: a page whose canonical link points at a different built page on siteUrl is a duplicate by design (a mirror, a print version, another rendering of the same content). It is left out of the duplicate check, is not required in the sitemap and is not reported as an orphan; listing it in the sitemap is a warning (non-canonical page in the sitemap), since sitemaps should list canonical URLs only.

Site-wide

  • robots.txt (seo.robots): no robots.txt in the build root is a warning. A Disallow: / in a group that applies to User-agent: * is an error (robots.txt disallows everything for User-agent: *). No Sitemap: line is a warning.
  • Sitemap (seo.sitemap): vidimus reads /sitemap.xml, /sitemap-index.xml, /sitemap_index.xml and every Sitemap: URL from robots.txt that exists in the build, and follows <sitemap><loc> entries of sitemap indexes. Every <url><loc> must be absolute, on siteUrl (when set), and a built page; a listed noindex page is an error (noindex page in the sitemap). If no sitemap is found at all you get a no sitemap.xml warning; otherwise every indexable page missing from it is a warning (indexable page missing from the sitemap). <![CDATA[…]]> and entities in <loc> are handled.
  • Orphans (seo.orphans): an indexable page that no other built page links to with an <a href> is a warning (orphan page: no other page links to it). Links from pages skipped by seo.exclude, such as the 404 page, still count; self-links do not. The home page / is never an orphan.

Example output

─── seo ─────────────────────────────────────────────────────────

✖ noindex in the production build
    on: /en/search/ /fr/search/ /it/search/ +1 more
    → Remove the robots noindex meta tag, or add the path to seo.allowNoindex if it is intentional.

✖ hreflang not reciprocated
    /en/pricing/ → /it/prezzi/
    on: /en/pricing/
    → Add a matching <link rel="alternate" hreflang> back to the linking page on each target page.

⚠ title outside 10-60 characters
    /blog/a-very-long-post/: 74
    on: /blog/a-very-long-post/
    → Rewrite the <title> to 10-60 characters, or adjust seo.titleLength.

⚠ robots.txt has no Sitemap line
    in: dist/robots.txt
    → Add a line like "Sitemap: https://example.com/sitemap.xml" to robots.txt.

⚠ orphan page: no other page links to it
    on: /legal/cookies/
    → Link to the page from navigation or another page, add it to seo.exclude if it is intentionally unlinked, or turn seo.orphans off.

✖ seo: 42 pages, 5 problem(s) (0.3s)

Options

Key Default Description
seo.exclude ['^/404(\\.html|/)?$'] URL path patterns to skip
seo.allowNoindex [] URL path patterns allowed to be noindex (search, thank-you pages)
seo.titleLength { min: 10, max: 60 } warn when the title length is outside the range
seo.descriptionLength { min: 50, max: 160 } warn when the description length is outside the range
seo.canonical true check canonical links
seo.h1 true warn on zero or several <h1>
seo.sitemap true check the sitemap and that indexable pages are in it
seo.robots true check robots.txt
seo.orphans true warn about pages nothing links to

Top-level keys used: siteUrl, exclude. The robots.txt file is still read for Sitemap: lines when seo.robots is off.

import { defineConfig } from 'vidimus';

export default defineConfig({
  siteUrl: 'https://example.com',
  audits: ['i18n', 'csp', 'a11y', 'links', 'r12s', 'seo'],
  seo: {
    allowNoindex: ['^/(en|it)/search/$', '/thanks/$'],
    titleLength: { min: 10, max: 70 },
    orphans: false,
  },
});

Common fixes

A complete <head> for a page with one translation:

<html lang="en">
  <head>
    <title>Pricing and plans | Example</title>
    <meta name="description" content="Compare the Free, Team and Business plans and see what each includes.">
    <link rel="canonical" href="https://example.com/en/pricing/">
    <link rel="alternate" hreflang="en" href="https://example.com/en/pricing/">
    <link rel="alternate" hreflang="it" href="https://example.com/it/prezzi/">
    <link rel="alternate" hreflang="x-default" href="https://example.com/en/pricing/">
  </head>
</html>

The Italian page carries the same three alternate links and its own canonical, https://example.com/it/prezzi/. Each page an hreflang link targets must link back to the page that names it, or the pair fails with hreflang not reciprocated.

A robots.txt that allows crawling and points at the sitemap:

User-agent: *
Disallow:

Sitemap: https://example.com/sitemap-index.xml

Keep a noindex page out of the sitemap and out of the findings: exclude it in the sitemap generator (for @astrojs/sitemap, the filter option) and add its path to seo.allowNoindex. For a page that should stay unlinked, such as a campaign landing page, add it to seo.exclude instead of turning seo.orphans off.

Staging builds often add noindex on purpose. Audit a production build. See Frameworks for per-framework notes and Adopting for rolling out seo on a large site.

Edit this page on GitHub