Skip to content
Learn

International SEO and special search

In this lesson

Section titled “In this lesson”
  • Choose a URL structure for a site in many countries: ccTLD, subdomain or subfolder.
  • Write hreflang annotations that Google accepts, and find the common mistakes.
  • Make a video eligible for video results with a watch page, a thumbnail and metadata.
  • Explain how an online store and a news site get their content into Shopping, Top stories and Google News.
  • Prepare an article for Google Discover, and say why Discover traffic is hard to plan.
Section titled “Two kinds of special search”

Day 09 was about place: a searcher near a business. This lesson has two other cases where normal SEO is not enough.

  1. International SEO: one site serves people in many languages or countries. The question is: which version does each searcher see?
  2. Special search surfaces: video results, shopping results, news results and the Discover feed. Each surface has its own entry rules on top of normal SEO.

Most sites need only some of this. A store in one country can skip hreflang. A blog without video can skip video SEO. Read the sections that match your site, but know the shape of all of them.

What matters most? For international SEO, the URL structure is the big decision, because a change later means a site migration (day 09c). Hreflang is where most errors happen. For the special surfaces, the rule is the same each time: the page must first be indexed and helpful, and then the surface adds its own requirements.

International SEO

Section titled “International SEO”

International SEO is the work that makes a site serve and rank for people in more than one country or language. Google separates two cases:

  • A multilingual site has content in more than one language. Example: a Canadian business with English and French pages.
  • A multi-regional site targets more than one country. Example: a maker that ships to Canada and the United States, with different prices.

A site can be both. Google tries to match the language of the searcher for the first case, and the locale for the second. A locale is a language plus a country, for example English for the UK.

Geo-targeting and localization

Section titled “Geo-targeting and localization”

Geo-targeting means that you tell search engines which country a page is for. Localization means that you adapt the page for that country: language, currency, prices, addresses, phone numbers, units and examples. A translation alone is not a full localization.

Google lists the signals that it uses to find the target locale of a page:

Signal What Google says
ccTLD A strong signal for a country, for example .de
hreflang Language and region labels, in HTML, HTTP headers or sitemaps
Server location A weak signal, because many sites use a CDN
Other signals Local addresses and phone numbers, local language and currency, links from local sites, the Business Profile

Google also lists what it ignores. It ignores location meta tags such as geo.position. It does not use the lang attribute or the URL to detect the language of a page. It reads the visible text. So use one language on each page, for both the content and the navigation. Do not put two translations side by side.

Search Console had a country-targeting setting for many years. Google removed it in 2022. Its help page says that the setting had “little value for the ecosystem”. Hreflang is still supported.

Bing is different, but its own guidance did not load in this run. Secondary sources say that Bing uses the content-language meta tag and the lang attribute more than hreflang. This is not confirmed, so set a correct lang attribute on every page too. It costs nothing.

Do not adapt by IP address

Section titled “Do not adapt by IP address”

Some sites detect the visitor’s country from the IP address and change the content or redirect. Google advises against this for three reasons:

  • IP location is often wrong.
  • Googlebot crawls mostly from US addresses. It does not set the Accept-Language header. So Google can miss the other versions.
  • An automatic redirect can block people and crawlers from the version that they want.

Give each version its own URL. Add links to the other versions, for example a language and country selector. A banner that suggests another version is fine. A forced redirect is not.

ccTLD, subdomain or subfolder

Section titled “ccTLD, subdomain or subfolder”

A ccTLD (country-code top-level domain) is a domain ending for one country, such as .de or .fr. A gTLD (generic top-level domain) has no country, such as .com or .org. Google compares four URL structures:

Structure Example Pros Cons
ccTLD example.de Clear geo-targeting; server location does not matter; easy to separate Expensive; more infrastructure; rules on who can buy some ccTLDs; one country only
Subdomain de.example.com Easy to set up; servers can be in different places Users cannot tell if de means a language or a country
Subfolder example.com/de/ Easy to set up; low maintenance on one host Users cannot tell the target from the URL; harder to separate
URL parameter example.com?loc=de — Not recommended

Two details surprise many people. First, Google treats some ccTLDs as generic: .io, .ai, .co, .me, .tv and others. So a .io domain does not target the British Indian Ocean Territory. Second, Google treats .eu and .asia as generic too.

Which structure is best? Google’s table lists pros and cons, not a winner. The choice is about cost, signals and teams. A subfolder is the cheapest to run, because everything stays on one host. A ccTLD gives the clearest country signal, but each domain is a separate site to run. A subdomain is between the two. Pick one structure early and keep it. A change later is a migration, with redirects for every page.

Hreflang

Section titled “Hreflang”

Hreflang is an annotation that tells Google about the language and region versions of a page. Google then shows the best version to each searcher. The roadmap calls it an HTML attribute. It can also go in an HTTP header or in an XML sitemap. Google says that the three methods are equal, so use only one.

The HTML form is a set of <link> elements in the <head>:

<link rel="alternate" hreflang="en-GB" href="https://example.com/en-gb/" />
<link rel="alternate" hreflang="en-US" href="https://example.com/en-us/" />
<link rel="alternate" hreflang="de" href="https://example.com/de/" />
<link rel="alternate" hreflang="x-default" href="https://example.com/" />

The rules from Google’s docs:

  • List every version, including the page itself. Every version carries the same set of links.
  • Return links are required. If page A lists page B, page B must list page A. If not, Google ignores the pair. This stops other sites from claiming to be your translation.
  • Use full URLs with https://.
  • Codes are language first, then an optional region. The language is ISO 639-1 (en, de). The region is ISO 3166-1 alpha-2 (GB, CH). A region alone is not valid. be means Belarusian, not Belgium.
  • Some codes do nothing. en-UK has no effect, because the code for the United Kingdom is GB. es-419 (Spanish for Latin America) is not supported.
  • Use x-default for everyone else. It names the fallback page, often a language selector.
  • Add a language-only page when you have several regions of one language. An en page catches English speakers in countries without their own version.

Hreflang also answers a duplicate content question from day 05. Translated pages are not duplicates. Pages in the same language for different countries can look like duplicates. For those, Google says to use hreflang together with a self-referencing canonical on each version. You learned canonicals on day 03.

Which pages list all the versions, including the return links?

Toy model with example data. Choose each searcher and see which version Google shows. Then untick the en-GB page: the UK searcher gets the wrong page, because a missing return link cancels the pair. Untick x-default and notice that the French searcher has no match at all.

Video SEO

Section titled “Video SEO”

Video SEO is the work that helps videos appear in search: in the main results, the Videos tab, Google Images, Discover and on YouTube. You saw video packs on day 04 and video structured data on day 05b. This section is about the conditions for those features.

Google’s first rule is a watch page: a page whose main purpose is to show one video. A video landing page and a sports highlight page are watch pages. A blog post that reviews an embedded video is not, and neither is a product page with a 360° video. Only watch pages are eligible for the full video features. Other pages can still show as a text result with a video badge.

Google needs these things to index a video:

  • The watch page is indexed and performs well in Search. An indexed page does not always mean an indexed video.
  • The video is in the HTML. Use <video>, <iframe>, <embed> or <object>. If JavaScript adds the player, check the rendered HTML (day 03b). Do not load the video only after a click or a swipe.
  • A valid thumbnail at a stable URL. Some CDNs make URLs that expire fast. Then Google cannot index the video.
  • Metadata. Use VideoObject structured data, a video sitemap or Open Graph tags. Keep the title, description and thumbnail unique for each video.

A video has three URLs, and people mix them up: the watch page, the player (embedUrl) and the video file (contentUrl). Google needs to fetch the video file for two features. Video previews show a few seconds of the video. Key moments link to parts of the video, like chapters. You can mark key moments yourself with Clip or SeekToAction markup, or with timestamps in a YouTube description.

YouTube is a search engine of its own. YouTube says that its search uses three elements: relevance (title, tags, description and content), engagement (for example, watch time for a query) and quality (expertise, authority and trust of the channel). If you embed a YouTube video on your site, Google can index both the YouTube page and your page.

E-commerce SEO

Section titled “E-commerce SEO”

E-commerce SEO is SEO for online stores. The main pages are category pages and product pages. Stores have three typical problems: many URLs, fast-changing data and thin pages.

Many URLs. Filters, sorting and variants make a huge number of URLs. You learned faceted navigation and URL parameters on day 03. Google adds these points for stores:

  • Give each product variant (one size and colour) its own URL, for example /t-shirt/green or /t-shirt?color=green. Do not use #green, because Google ignores fragments.
  • If an optional parameter marks the variant, make the URL without it the canonical.
  • Use ?key=value parameters, not ?value.
  • Do not link to session IDs or tracking parameters.

Crawl paths. Googlebot does not type into a search box. So link every product from a category page with normal <a href> links. If that is not possible, use a sitemap or a Merchant Center feed. Link your best sellers from the home page, because more internal links show more importance. If a category is empty, use noindex, or return 404 if you remove it.

Product data. Google gives two ways to share product data, and recommends both:

Method What it does
Product structured data Makes product pages eligible for rich results with price, availability, ratings and shipping. Use “merchant listing” markup when people can buy on the page, and “product snippet” markup for review pages.
Google Merchant Center A feed of your products. It is required for the Shopping tab. It updates faster than crawling, and it can share data that is not on the site, such as stock in each store.

Keep the two in sync. If a product sells out on the site but the feed still says “in stock”, Google sees a conflict. Merchant Center has an automatic update setting for this.

Content. Google’s store guide ends with one rule: create “useful and interesting content that is valuable to users”. Product text that is the same on many stores adds nothing new. Add what only you have: real photos, sizing advice, comparisons and honest reviews. This is information gain from day 05.

News SEO

Section titled “News SEO”

News SEO is the work that helps a publisher appear in news surfaces. Google has three: Google News (the app and news.google.com), Top stories (a carousel in the main results) and the News tab in Search.

There is no application. Google retired the old application process in December 2019. Google says that every site is “automatically considered”. To be eligible, a site needs:

  • High levels of expertise, authority and trust.
  • A consistent history of original news content.
  • Compliance with Google’s News policies.

Google then ranks news content by relevance, prominence, authority, freshness, location and language.

Publisher Center used to let publishers set up a publication page. In 2024, Google stopped this, and said that the pages would be made automatically from early 2025. Google says that Publisher Center does not change eligibility.

A news sitemap helps Google find new articles fast. List only articles from the last two days. One news sitemap can have up to 1,000 <news:news> entries. Do not make a new sitemap for each update; update the same file.

To measure, use the Google News report and the News search type in Search Console. You saw search types on day 08.

Google Discover

Section titled “Google Discover”

Google Discover is a feed of content that is part of Google Search, mostly on mobile devices. It shows content that matches each person’s interests, based on their Web and App Activity. There is no query. The roadmap puts it next to news, but Discover shows all kinds of content: articles, videos, sports, health and lifestyle. Older content can also appear.

Every indexed page that follows Discover’s content policies is eligible. No special tag is needed. Eligible does not mean that it will appear.

Google’s advice has four parts:

  • Titles and headlines that give the essence of the content. No clickbait: a preview that misleads, exaggerates or withholds the key information to get clicks.
  • No sensationalism that plays on outrage or morbid curiosity.
  • Timely, well-told, original content.
  • Large, relevant images. At least 1200 px wide, more than 300,000 pixels, ideally 16:9. Large images need max-image-preview:large in the robots meta tag, or AMP. Name the image in og:image or in structured data. Do not use a logo.

On 5 February 2026, Google released the first Discover core update. It showed more local content from sites in the user’s country, and less sensational content and clickbait. It also showed more content from sites with expertise in the topic, judged topic by topic. It started for English-language users in the US. Google said that it would expand to all countries and languages later.

Discover traffic is less predictable than search traffic. Interests change, Google changes the content types, and core updates move traffic. Google says to treat Discover as extra traffic, not as the base of a plan. Track it with the Discover performance report in Search Console. It shows 16 months of impressions, clicks and CTR.

Toy model of Google's Discover image and headline advice. Move the slider until the card turns large, then switch the robots meta tag. Notice that one wrong setting is enough to drop to a small thumbnail.

Key terms

Section titled “Key terms”
  • International SEO: the work that makes a site serve and rank for people in more than one country or language.
  • Multilingual site: a site with content in more than one language.
  • Multi-regional site: a site that targets more than one country.
  • Locale: a language plus a country, for example English for the UK.
  • Geo-targeting: telling search engines which country a page is for.
  • Localization: adapting a page for a country: language, currency, prices, addresses and examples.
  • ccTLD: a domain ending for one country, such as .de.
  • gTLD: a domain ending with no country, such as .com; Google also treats some ccTLDs, such as .io, as generic.
  • Hreflang: an annotation that tells Google the language and region versions of a page.
  • Return link: the hreflang link back from page B to page A; without it, Google ignores the pair.
  • x-default: the hreflang value for the fallback page for all unmatched searchers.
  • Watch page: a page whose main purpose is to show one video.
  • Key moments: links to parts of a video in search results, like chapters.
  • Product variant: one combination of product attributes, such as a size and a colour.
  • Google Merchant Center: Google’s service for product feeds; it is required for the Shopping tab.
  • News sitemap: a sitemap of articles from the last two days, for Google News.
  • Google Discover: a feed of content based on a person’s interests, with no query.
  • Clickbait: a title or image that misleads, exaggerates or withholds key information to get clicks.

Exercise

Section titled “Exercise”

Exercise · about 15 min

Check the hreflang and Discover setup of a news site

You need only a browser. Use the BBC, which has a UK site and an international site.

  1. Open view-source:https://www.bbc.co.uk/news in Chrome or Firefox. Search the source for hreflang.
  2. Write down each hreflang code and its URL.
  3. Open view-source:https://www.bbc.com/news. Search for hreflang again. Check the return links: does each page list the other and itself?
  4. Check for an x-default link. Who gets which version?
  5. On both pages, search for max-image-preview and og:image. Compare them with Google’s Discover image advice.
  6. Open one article from the page. Repeat step 5 for the article.
  7. Write one sentence for each page: is its setup correct, and what would you change?
Show a possible solution

Results on 21 October 2026. A site can change, so your results can differ.

  • bbc.co.uk/news lists en → https://www.bbc.com/news and en-gb → https://www.bbc.co.uk/news. It lists itself, and it has a self-referencing canonical.
  • bbc.com/news lists the same two links. So the return links are correct.
  • There is no x-default. That is fine here: the en page is the fallback for English speakers outside the UK. Speakers of other languages also get a version that Google chooses.
  • bbc.co.uk/news has max-image-preview:large. Its og:image file name says 1024 × 576, which is below the 1200 px width that Google advises. It is a section page, so this matters less than on an article.
  • Both pages have lang="en-GB" on the <html> element. Google ignores this for language detection, but other search engines can use it.

A good conclusion: the hreflang pair is correct. On articles, check that the image is at least 1200 px wide and that large previews are allowed.

Check yourself

Section titled “Check yourself”
A shop has example.com/en-gb/ and example.com/en-ie/. Both pages list each other in hreflang. The Irish page lists the UK page, but the UK page does not list the Irish page. What happens?
A start-up uses a .io domain and wants to target Germany. Which statement is correct?
A recipe blog embeds a cooking video in each long article. Why does Google not show these pages in the Videos tab?
A store sells a shirt in five colours. Today it shows each colour with /shirt#red, /shirt#blue and so on. What is the problem?
A publisher's Discover clicks fall by half in one week. Its search traffic does not change. Which first step fits Google's guidance?

Sources

Section titled “Sources”

Next

Section titled “Next”

Day 09c is the last SEO day: it covers site migrations, redirect mapping, the checks after a move, and how to run a full SEO audit.