usdigitalmarketingindex.com SEO
Should international XML sitemaps include regional pages that have not yet been translated?
Updated November 15, 2026
Include every page that exists, is published, and is meant to be indexed, whatever language it is in. Leave out pages that do not exist yet. An untranslated page is one of two things, and the answer differs for each. If a regional URL is live and shows real content, even content still in the source language, it is a real page and can be listed like any other. If “not yet translated” means the regional URL has not been created, there is nothing to list, and nothing to name as an alternate. The sitemap describes the site as it is today. The same holds for hreflang entries carried in the sitemap: Google’s documentation requires each entry to list “every language/locale variant of the page including itself”, and a variant is a page that exists (Google’s documentation on localized versions of pages).
Separating published from planned pages
Keep an inventory that states, for every piece of content and every market, what exists.
| State | Meaning | Lives at a URL? |
|---|---|---|
| Published, localized | Translated and adapted, live | Yes |
| Published, source-language fallback | Live at the regional URL, showing source-language content | Yes |
| In translation | Being worked on, not live | No |
| Planned | On the roadmap | No |
| Not planned | This content will not exist for this market | No |
| Retired | Removed | No longer |
An inventory of localized URLs for a hypothetical site with three markets:
| Content | en-GB | de-DE | fr-FR |
|---|---|---|---|
| Home | /en-gb/ published | /de-de/ published | /fr-fr/ published |
| Boiler cover | /en-gb/boiler-cover/ published | /de-de/kesselschutz/ published | In translation; no URL |
| Pricing | /en-gb/pricing/ published | /de-de/preise/ published | /fr-fr/tarifs/ published |
| Winter guide | /en-gb/guides/winter/ published | Planned; no URL | Not planned |
| Legal notice | /en-gb/legal/ published | /de-de/impressum/ published | /fr-fr/mentions-legales/ published |
From this table, the sitemap contains the published URLs: five for en-GB, four for de-DE, and three for fr-FR, twelve in all. The French boiler cover page and the German winter guide are absent because they are not pages.
Avoiding nonexistent hreflang targets
The rule for alternates follows from the inventory. An hreflang entry names a URL for a language and region. If that URL does not exist, the entry points at nothing.
For the boiler cover page, the alternates today are:
<url>
<loc>https://www.example.com/en-gb/boiler-cover/</loc>
<xhtml:link rel="alternate" hreflang="en-GB" href="https://www.example.com/en-gb/boiler-cover/"/>
<xhtml:link rel="alternate" hreflang="de-DE" href="https://www.example.com/de-de/kesselschutz/"/>
</url>
<url>
<loc>https://www.example.com/de-de/kesselschutz/</loc>
<xhtml:link rel="alternate" hreflang="en-GB" href="https://www.example.com/en-gb/boiler-cover/"/>
<xhtml:link rel="alternate" hreflang="de-DE" href="https://www.example.com/de-de/kesselschutz/"/>
</url>
The two entries are an excerpt, not a complete file. They sit inside the sitemap’s urlset element, which must declare the xhtml namespace for the xhtml:link lines to be valid:
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"
xmlns:xhtml="http://www.w3.org/1999/xhtml">
There is no fr-FR line. It is added, to both entries and with an entry of its own, on the day the French page is published.
What not to do:
| Practice | Problem |
|---|---|
| Listing the planned French URL as an alternate | The target returns 404, or a redirect; the annotation is false |
| Pointing the fr-FR alternate at the English page | It tells Google the English page is the French version |
| Pointing every missing alternate at the home page | The home page is not the equivalent of an inner page |
| Publishing an empty or placeholder page so the URL exists | A page without content, listed for indexing |
| Publishing unreviewed machine translation to fill the gap | Content the business has not checked, presented as its localized page |
A cluster may be incomplete. The documentation allows for sets that do not cover every language, noting that Google will still process the alternates that point to each other.
The fallback case needs a deliberate decision. Suppose the French site shows the English boiler cover page at a French URL until translation is done. That page exists and can be listed. Consider, though, what the documentation says about duplicates: localized versions are treated as duplicates only “if the main content of the page remains untranslated”. An English page at a French URL is that situation. Options are to wait and publish the URL when the translation is ready, or to publish the fallback knowingly, with the navigation making clear what the visitor is getting. Declaring it fr-FR in hreflang while its content is English misdescribes it.
Sitemap completeness
| Include | Exclude |
|---|---|
| Every published, indexable URL in every market | URLs that return 404 or redirect |
| The canonical form of each URL | Non-canonical duplicates and parameter variants |
| Pages in the source language that are live for a market | Planned and in-progress pages |
| Each entry's real alternates, including itself | Alternates that do not exist |
| Pages set to <!–INLINECODE2–> |
Organization for an international site:
- One sitemap per market or language keeps the files readable and lets Search Console report on each.
- A sitemap index lists them.
- If hreflang is carried in sitemaps, every URL in a cluster needs its own entry with the full set of child entries. The documentation’s example: three versions of a page mean three entries, each with three identical child entries.
- Generate sitemaps from the inventory or the publishing system, so a page appears when it is published and disappears when it is retired.
A sitemap is a list of URLs offered for crawling. Inclusion does not cause indexing, and omission does not prevent it. Its value here is accuracy: it is a statement of which localized pages exist.
Release verification
When a translation is published:
| Step | Check |
|---|---|
| 1 | The new URL returns 200 and shows the translated content |
| 2 | Its canonical tag is self-referencing |
| 3 | It appears in the market's sitemap |
| 4 | Its hreflang set lists itself and every existing version |
| 5 | Every existing version's set now lists the new URL |
| 6 | The inventory shows it as published |
| 7 | Language switcher links on the other versions point to it |
| 8 | The sitemap has been regenerated and is fetchable |
| 9 | Search Console shows the sitemap read without errors |
Step 5 is easy to miss, because it changes pages other than the one being released. Generating all sets from the inventory makes it automatic.
When a localized page is retired, run the list in reverse: remove it from the sitemap and from every set that names it, and decide its redirect.
Run two periodic comparisons. First, the inventory’s published URLs against the sitemap’s URLs: any URL in one and not the other is an error to resolve. Second, the URLs named in hreflang against the inventory: every one must be a published page. The reverse does not hold. A published page with no other language version, such as the en-GB winter guide in the inventory above, belongs in the sitemap and needs no hreflang set.
Reciprocity between the versions that do exist is a separate check from the question of which versions exist, and it is covered in the guide to how missing reciprocal hreflang references can be detected.
The rule: list what is published, in whatever language it is published, and name as alternates only pages that are live today.
Add your comment to this answer