UNmiss Blog

IKEA Keeps 70 Hreflang Tags Out of Every Page

IKEA publishes 116 country and language versions and not one hreflang tag on a product page. They live in 2,172 sitemap files instead, one of which is 47.6 MB.

IKEA sells the same bookshelf in about 60 countries.

Somebody in Estonia searches for it in Estonian. Somebody else in Estonia searches in Russian. A third person searches in English, from Estonia, because they moved there last year. Three people, one product, three different pages that should win.

Multiply that by every product IKEA sells and you have one of the hardest technical problems in SEO: telling a search engine which version of a page belongs to which reader.

The usual answer is a tag in the page called hreflang. So we opened an IKEA product page and looked for it.

There isn't one. Not in the HTML, not in the response headers. IKEA moved the whole system somewhere else — and once you see where, the decision is obviously right.

Results at a glance

116
country and language versions
0
hreflang tags on a product page
2,172
sitemap files carrying them instead
The short version

IKEA declares its language alternates in its sitemaps, not in its pages. At around 70 alternates per product, putting them in the page head would add roughly 7 KB of tags to every product page on the site.

The weight does not disappear, though. It moves — and you can watch it pile up.

116 versions, 47 of them English

Start at the front door. IKEA's global homepage lists 116 alternate versions of itself, each one a country and language pair, at addresses shaped like /ee/et/ and /ee/ru/ — country first, then language.

Here is the part people get wrong about international SEO. 47 of those 116 are English.

Not 47 translations. The same language, offered to 47 different markets, because an English page for Canada is not an English page for Qatar: different prices, different currency, different stores, different stock.

IKEA is not running a translation project. It is running a market project that happens to involve translation, and the URL structure says so — the country comes first.

The tag that isn't on the page

We sampled twelve of the 116 versions and fetched each one. Eleven returned 200. None of them carried a single hreflang tag in the HTML, and none carried one in the HTTP headers either.

That looks like a serious mistake, and for most sites it would be. Hreflang has to be reciprocal: if page A claims B as its alternate, B has to claim A back, or search engines are entitled to ignore the whole cluster.

But there is a third place the rule allows you to declare it, and that is where IKEA put everything.

2,172 sitemaps, one per locale

IKEA's sitemap index points at 2,172 files. They are named by locale and by page type — products, categories, planners, series, store events, second-hand — so there is a file for products in Estonian-for-Estonia, another for categories in English-for-Belgium, and so on across 91 locales.

Inside those files, every single URL carries its own list of alternates.

This is the right call. Seventy <link rel="alternate"> tags in a page head is about 7 KB of markup on every product page, shipped to every visitor, on a site where most visitors are on a phone. Moving it into the sitemap means only the crawler pays.

The weight has to go somewhere

But it does not vanish, and this is the part worth taking away.

We measured one of those product sitemaps. It holds 4,526 products and 317,599 hreflang entries, and the file is 47.6 MB.

The sitemap protocol allows a maximum of 50 MB per file. IKEA is at 95% of the limit with fewer than five thousand products in the file.

So the alternates are not free. They set the size of every sitemap file, and they cap how many products can go in one. That is why there are 2,172 files rather than a couple of dozen: the hreflang, not the product count, is what forces the split.

What our tool found inside

We then read one of IKEA's smaller sitemaps — the Estonian planner file — with our own XML Sitemap Generator, which checks a published sitemap against the protocol without visiting any of the URLs in it.

It came back clean on the basics: 48 URLs, 389.5 KB, HTTP 200, and every entry carrying a readable lastmod. Then it flagged two things.

Every URL in the file sits above the sitemap's own folder. The file lives at /sitemaps/ while the pages live at /ee/et/planners/. A sitemap is only supposed to cover its own path and below. Search engines are usually forgiving about this when the sitemap is submitted directly, but it is a rule IKEA is technically breaking on every one of those 2,172 files, and moving the files to the site root would fix the whole class of problem.

Every entry carries priority and changefreq. Google ignores both fields outright. They are valid, and they do nothing.

Which is a small thing on most websites and not a small thing here. Those two ignored fields are being written into a file that is already at 95% of its maximum size. On the product sitemaps, they are spending headroom on data nobody reads.

UNmiss Backlink Analyzer report for ikea.com showing Domain Rank 93 out of 100, about 14.8 million total backlinks, 324,311 referring domains and 81 percent dofollow links.
What to notice: 324,311 referring domains. Whatever the architecture costs to maintain, it is holding up an enormous amount of authority across 91 markets.

Copy this in an afternoon

You do not have 91 locales. Four steps anyway.

1. Count your alternates before you choose where to put them. Under about ten, the page head is fine and simpler to debug. Above that, the markup starts to cost real bytes on every page, and the sitemap becomes the better home.

2. Pick one method and use it everywhere. Head tags, HTTP headers or sitemap — all three are valid, and mixing them is how clusters end up half-declared and quietly ignored.

3. Split sitemaps by locale, not just by section. IKEA's naming means a change in one market regenerates one market's files. If your sitemaps are organised by page type alone, every locale change rewrites everything.

4. Drop priority and changefreq. Google ignores them. They cost you bytes in a file that has a hard size limit, and they buy nothing.

Then read your own sitemap back and see what you actually published, rather than what you meant to.

What you cannot copy, and what we missed

The operational machine underneath. Keeping 91 locales in sync is a supply-chain problem before it is an SEO one — prices, stock and store data all have to be right per market. The sitemap architecture is the visible end of something much larger.

And here is what we could not check:

We sampled, we did not crawl. Twelve of 116 country pages, four of 2,172 sitemap files. The pattern was consistent across everything we opened, but we did not read all of it.

We could not read IKEA's biggest sitemap with our own tool. At 47.6 MB it exceeds the 5 MB our free sitemap reader accepts, so the numbers for that file come from fetching and counting it directly instead. We would rather tell you that than quietly use a smaller file and imply otherwise.

We cannot see what Google does with any of it. Declaring an alternate is a request. Whether search engines honour every one of those 317,599 entries is not visible from outside.

No traffic or revenue figures. None were available and none are claimed.

Free, no account
Read your own sitemap back

Paste the address of a sitemap you have already published and our XML Sitemap Generator checks it against the protocol without visiting a single URL in it — size, entry count, lastmod, and the mistakes that quietly stop it working. It is the same check that found the two issues in IKEA's file above.

  • Flags URLs outside the sitemap's own path
  • Points out fields Google ignores, like priority and changefreq
  • Hands back a cleaned-up copy you can publish
Check a sitemap free

Frequently asked questions

How many versions of itself does IKEA publish?

Its global homepage declared 116 alternates on 22 August 2026, plus one x-default, each a distinct country and language pair at URLs shaped /{country}/{language}/. The sitemap filenames reveal 91 distinct locales, so not every homepage alternate has a full sitemap set behind it.

Why are 47 of the 116 in English?

Because they target different markets rather than different languages. English for Canada, Ireland, the Netherlands, Norway, Qatar and dozens more are separate pages with separate prices, currencies, stock and stores. The language is the same; the commercial reality is not, which is why the country comes first in the URL.

Is it a problem that IKEA's pages have no hreflang tags?

No. The specification allows you to declare alternates in the page head, in HTTP headers, or in the XML sitemap, and IKEA uses the third. We checked the first two on a sample of twelve country pages and found nothing, which is consistent rather than broken — a site should pick one method and stick to it.

How big are these sitemap files?

The Estonian product file holds 4,526 products with 317,599 hreflang entries and weighs 47.6 MB, against a protocol maximum of 50 MB. Others we sampled ran at roughly 8 to 11 KB per URL: a categories file at 1,057 URLs and 94,486 alternates, a series file at 345 URLs and 30,792 alternates, and a planner file at 48 URLs and 3,620 alternates. Some are empty, at 276 bytes with no URLs at all.

What did your sitemap tool find in IKEA's file?

Reading the Estonian planner sitemap, it reported 48 URLs at 389.5 KB, HTTP 200, with a readable lastmod on every entry. It raised two issues: all 48 URLs sit above the sitemap's own folder, since the file is at /sitemaps/ while the pages are at /ee/et/planners/; and all 48 carry priority or changefreq, which Google ignores.

Should I put hreflang in my pages or my sitemap?

Count your alternates. Below roughly ten, the page head is easier to inspect and debug. Above that, the tags add real weight to every page — at IKEA's 70 alternates it would be around 7 KB per product page — and the sitemap is the better home. What matters most is choosing one method and applying it everywhere, because mixed implementations are how clusters end up ignored.

What is IKEA's backlink profile?

Our Backlink Analyzer returned Domain Rank 93 out of 100, 324,311 referring domains, 12,475,410 dofollow and 2,924,451 nofollow links, 139,630 referring IPs and 64,336 referring subnets on 22 August 2026.

Can I check this myself?

Yes. Open any large international site's robots.txt, follow the sitemap, and look at whether the URL entries contain xhtml:link alternates. Then view the source of one of its pages and search for hreflang. Where the tags live tells you a great deal about how seriously the site takes international search.

Decide where your hreflang lives

Every company in this series built something. RTINGS measures what nobody else measures. Zapier publishes a fraction of what it could. Zillow never deletes a sold home. Calculator.net stopped at 222 pages. wikiHow draws a picture for every step. NerdWallet puts three names on every article.

IKEA's contribution is quieter and more technical: it decided that the instructions for search engines do not belong in the page.

That sounds like a small architectural preference. At 116 versions it is the difference between shipping seven kilobytes of link tags to every shopper on earth and shipping none of it — paid for instead in sitemap files that sit, deliberately, just under the line.

So look at your own pages and ask what is in the head that no human will ever read, and whether it needs to be there at all.

Then read your sitemap back and find out what you have actually been publishing.

Measured on 22 August 2026. Homepage alternates were counted from the rendered HTML; sitemap counts and file sizes come from fetching the files directly and counting entries. Page-level checks are a sample of 12 of 116 country versions and 4 of 2,172 sitemap files.

Link figures are a free run of our own tool and the sitemap findings come from our XML Sitemap Generator; exact values are in the FAQ. No traffic or revenue data for IKEA was available to us and none is claimed here.

Blog