Hreflang errors are the most expensive bug in iGaming SEO

An hreflang cluster only works if every page in it points at every other page, and every one of those pages points back. Miss one return link and the whole cluster is discarded, not just the broken pair. That is why the failure is all-or-nothing and why it is so easy to miss.

Why iGaming gets this wrong more than most

Which hreflang errors break an iGaming cluster

/Markets launch one at a time, and each launch has to update every existing page in the cluster. By the fifth market that is four sets of edits nobody scheduled.
/Portuguese is two locales, not one. pt-BR and pt-PT are separate markets with separate regulators, and treating them as one is the most common single error we see.
/Templated bonus and game pages generate hreflang programmatically, so one bad rule produces thousands of broken URLs rather than one.
/Markets get pulled. When a jurisdiction closes and pages come down, the return links from every other locale usually stay pointing at them.

The four errors that actually break clusters

Cluster-breaking hreflang errors, in the order we find them
Error	What it looks like	What it costs
Missing return link	/de/ points at /en/, /en/ points at nothing	The entire cluster is ignored, including the pages that were correct
Pointing at a redirect or a 404	Market closed, pages removed, links left in place	Same as above, and it compounds every time a market changes
Wrong region code	pt for Brazil, or en-UK instead of en-GB	The tag is invalid and dropped. en-UK is the single most common typo
No self-reference	Each page lists its alternates but not itself	Cluster discarded. Easy to miss because the tags look complete

Ordered by frequency across iGaming audits, 2026.

There is a fifth that is not an error but behaves like one: x-default pointing at whichever page is being rendered. Every page then claims to be the fallback, and the signal is worthless. It should resolve to one fixed page for the whole group.

Why nothing warns you

Search Console retired its International Targeting report, which is where hreflang errors used to surface. There is no replacement panel. The tags either work or are silently discarded, and a discarded cluster produces no message, no coverage error and no ranking drop in your home market.

What you see instead is the newest market underperforming for reasons that look like content or competition. It gets treated as a market problem, more content gets written, and the actual fault is four lines in a template.

How to check in ten minutes

/Pick one page that exists in every locale. Fetch each version and list its hreflang tags.
/Check every tag resolves to a live URL with a 200, not a redirect. A redirect breaks the cluster as surely as a 404.
/Check each page includes itself. Self-reference is the one people leave out.
/Check the codes against the ISO lists. en-GB not en-UK, pt-BR and pt-PT as separate entries, and no bare pt.
/Check x-default appears once and points at the same URL from every page in the group.

If any of the five fails, the cluster is not being used, and it has probably not been used since the market that broke it launched. Fixing it is a template change; the recovery takes as long as a recrawl of every affected page, which on a large casino site is not fast.

How we handle this at scale is on the multilingual SEO page, and the reason Brazilian Portuguese gets its own research rather than a translation is set out on the Brazil market page.

Sources for the figures here: the 60% error rate comes from RedClaw's audit of more than 50 iGaming sites, 2026, and matches what we see across 40 markets. Wider international-site error rates of around 75% are reported by Digital Applied. Search Console's International Targeting report was retired in 2022 and has not been replaced.

Why templated sites make it worse

On a brochure site, hreflang is a handful of pages and a person can eyeball it. On a casino with several thousand game pages across five locales, the tags are generated, and one bad rule produces tens of thousands of broken declarations at once.

/A game available in three markets but not the fourth still gets a tag pointing at the fourth, which 404s.
/Provider pages and category pages often use different generation rules, so one set is correct and the other is not.
/Filtered and faceted URLs inherit hreflang from their parent, pointing at parameters that do not exist in the other locale.
/Promotional pages expire. When a bonus ends in one market and not another, the cluster breaks in exactly the pages that were converting.

The pattern is consistent: the pages that break first are the ones with the most commercial value, because they are the ones that change most often.

Fix it in this order

/Self-reference first. It is one line in the template and it is missing more often than any other element.
/Then the return links, starting from the most recently launched locale, because that is where the break almost always originates.
/Then the codes. en-GB not en-UK, pt-BR and pt-PT separately, and no bare pt or bare en.
/Then x-default, which should resolve to one fixed page for the whole group rather than to whichever page is rendering.
/Only then the long tail of expired and market-specific pages, which is the largest job and the least urgent.

Doing it in that order means the cluster starts working again after step two on most sites, and the remaining steps improve accuracy rather than restore function. Doing it in the opposite order, which is what a crawl report tends to suggest, means fixing thousands of low-value URLs while the cluster stays broken.

Related on this site

/Multilingual iGaming SEO — hreflang at scale, and why keyword research is redone per language.
/Technical SEO — indexation ceilings, crawl budget and the rest of the containment problem.
/Brazil market data — why pt-BR is researched rather than translated from English.
/Germany market data — a second locale where the same cluster errors show up.
/iGaming SEO — the full programme this sits inside.
/Pricing — what a multi-market engagement costs, published.
/Get the report — hreflang integrity is checked in every one.
/Glossary — hreflang, canonical and the terms around them, defined.
/All analysis — five pieces, each built on a figure published somewhere and connected nowhere.

Questions

Does one broken link really break the whole cluster?

Yes, and that is the part people find hardest to believe. Hreflang is treated as a set of mutual declarations. If page A claims page B as an alternate and B does not claim A back, the search engine cannot confirm the relationship and discards the group rather than guessing which half to trust.

Can we use pt for both Brazil and Portugal?

You can, and it will cost you. Brazilian and European Portuguese differ in vocabulary, register and search behaviour, and the regulators are entirely separate. Using bare pt tells the search engine one page serves both, so one of the two markets gets a page written for the other.

Where does Search Console report these now?

It does not. The International Targeting report was retired and nothing replaced it. Invalid tags are discarded silently, which is why hreflang has to be checked deliberately rather than waited on. This is the main reason the error survives so long on large sites.

How long does recovery take after a fix?

As long as a recrawl of every page in the cluster, which on a casino site with thousands of game pages is weeks rather than days. The fix itself is usually a single template change, so the cost is almost entirely in the waiting rather than the work.

Should hreflang go in the sitemap or the head?

Either works and both are valid. The head is easier to verify by eye and easier to break in a template. The sitemap is harder to inspect and easier to keep correct programmatically. Pick one and never split a cluster across both, which produces conflicting declarations.

Does hreflang help with AI search too?

Indirectly. It does not tell a language model anything, but it decides which version of your page gets indexed for which market, and a page that is not indexed cannot be retrieved. On the unsearched half of AI answers it makes no difference at all.

Free · No obligation

We check this in every audit.

Index coverage, hreflang integrity and AI visibility across six surfaces, in one market, free.

Run my audit

Core SEO

AI search

Operators

Partners