In this episode of the International Web Effectiveness Podcast, Bill Scully and Bill Hunt review the latest insights from Google’s Search Off the Record podcast, focusing on the complexities of HREFLang implementation. They discuss key takeaways, including the challenges of correctly targeting markets with HREFLang tags and the potential future automation of these processes by Google.
Key Takeaways
- HREFLang does not tell Google the market you are targeting, it’s targeting the market and language. This is important to remember when using HREFLang, as it is not a substitute for other methods of targeting specific markets like IP redirects.
- HREFLang triggers a crawling dependency, which can help accelerate the indexing of new market pages. This makes HREFLang a valuable tool for SEO, especially for site migrations.
- Google may be able to automate HREFLang in the future. This would help to ensure that HREFLang is used correctly and consistently, and would also help to reduce the number of problems that arise from incorrect or missing HREFLang.
- Language emotion and country promotion can affect how your content is ranked. This is something to keep in mind when creating content for different markets.
- Aspirational HREFLang can lead to problems with Google indexing. This is when a large percentage of your pages are in a language other than what you are listing in your HREFLang.
- Using the correct HREFLang can help to improve your website’s ranking in Google. This is because it helps Google to understand the language and market of your website, and to serve the most relevant results to users.
Catch the Full Episode
YouTube: https://youtu.be/Kksxcrk_WWc?si=2UkpegaMnFB66-pl
Spotify: https://open.spotify.com/episode/2mxjKoskDSzK2C9oiE3JgC?si=_INoJhg3SAW88lHX-nEaVw
Transcript
Bill Scully: Welcome to Episode 6 of the International Web Effectiveness Podcast, where Bill Hunt and I, Bill Scully, focus on strategies and techniques to improve your web presence and manage global websites. In our last episode, we discussed the root causes of web project failures. Today, we’re diving deep into a recent episode of Google’s Search Off the Record podcast, where they talked about website internationalization and Hreflang.
Bill, you wrote a pretty in-depth review of that podcast. It must have taken a long time to put together. Honestly, it took me longer to read your post than it did to listen to the podcast! You made some great observations, and I don’t want them to go unnoticed. Can you explain why you wrote the review and walk us through your key takeaways?
Bill Hunt: Yeah, I mean, it was a lot to go through. The reason I dug into it so deeply was that there were some fascinating insights that mainstream industry sources didn’t really mention. They picked up on a few clickbait-style topics—like the idea that Google might automate or eliminate Hreflang—but that’s not exactly what was said.
The second reason I wrote the review was that people were reading those articles and then sending me questions. To answer them properly, I wanted to make sure I fully understood what was discussed. I read the transcript, listened to the podcast multiple times, and then wrote responses to clients and former clients who had questions about it.
One of my favorite takeaways from the podcast was when Gary Illyes said that the language of a page alone is not a signal for which country it targets. We’ve talked about this before—back in Episode 1—where we discussed common frustrations SEOs have with Hreflang. Many people don’t fully understand what Hreflang actually does. It’s not just about identifying the language of a page—it’s about indicating the specific market that language version is for.
A great example is Nike. They have seventy-eight websites, with around thirty-eight in English. Out of those, twenty-two use a dollar symbol. Gary pointed this out. When Google sees thirty-eight different English-language pages, it’s hard to determine which one belongs to which market. Martin Splitt added that, for many sites, the only noticeable differences might be the currency, VAT tax, or minor changes in product sizing. He even mentioned shipping but noted that shipping details usually appear further down the purchase funnel. The key takeaway here is that Google finds it difficult to determine the market just from the page language, and Hreflang is meant to bridge that gap.
Another major point was Gary’s statement that Hreflang triggers a crawling dependency. What that means is when Google encounters Hreflang tags—whether in the header or an XML sitemap—it prompts Google to validate those alternate pages by crawling them. This has huge implications. If you launch a new market version of your site, this mechanism helps accelerate the indexing process. It’s something we see all the time during site migrations—implementing Hreflang speeds up Google’s ability to process alternate pages.
Lastly, let’s talk about automation. Martin asked why Google doesn’t just use the HTML language tag, which is a W3C requirement and appears in every page’s header. Gary explained that the reason they can’t rely on it is that most of the time, it’s incorrect. He gave the example of Joomla—years ago, if you installed Joomla, the default setting was US English, and there was no way to override it. No matter which market you deployed it for, the language tag remained en-US.
We still see this problem today. If a French development agency installs WordPress for a US-based client, there’s a good chance the US English site will end up with fr-FR tags because that’s the default language pack that was installed. These incorrect tags make it impossible for Google to use them as a reliable signal.
Bill Scully: That’s a great breakdown, Bill. It’s clear that there’s still a lot of misunderstanding around Hreflang, and Google’s insights really reinforce the importance of getting it right. Thanks for going through all of that!
Bill Hunt: Absolutely. It’s a critical part of international SEO, and the more we understand it, the better we can optimize for global search visibility.
Bill Hunt:
They got into an interesting discussion. Gary even said that enterprises should be able to trust Hreflang because the investment to implement it is significant. The amount of effort required for an enterprise to do it properly is high, and I found that kind of hilarious because it’s a common question—why aren’t companies doing it correctly?
Unless you do it right—some do it right—exactly. That’s the point. If companies are investing the resources, why not ensure it’s done correctly?
Bill Scully:
Yeah, exactly. It’s not like this is some small thing that only a handful of companies deal with.
Bill Hunt:
Right. At the end of the day, Gary mentioned that, ten years ago, Google could detect a lot of these issues when he first started working on Hreflang. Now, with all the advancements in AI and intelligence, it should be easier to handle. He didn’t give specifics on how they might automate it, but in his mind, it seemed like something that could be streamlined. If they manage to do that, we’d likely avoid many of the current issues—incorrect Hreflang implementation or companies not using it at all, it was an interesting discussion.
Another key topic I want people to pay attention to is language emotion and country promotion. This ties into our previous discussions about ccTLDs—country-level domains—and how some of them have been commercialized. For example, the .ai domain is now widely used for AI projects, just like Columbia’s .co and Tuvalu’s .tv were commercialized. These countries essentially sold the rights, shifting their domains from country-specific designations to global or generic top-level domains.
Bill Scully:
Yeah, and that kind of shift changes how Google interprets those domains, right?
Bill Hunt:
Exactly. Another technical aspect—pretty geeky—is Google’s use of attribute grammar language. Essentially, it analyzes the language of a query, its context, and then serves results that best match. But then, personalization comes into play.
For example, if someone searches for “How do you spell banana?” that’s clearly an English-language query. But in some cases, users were receiving German results. That’s because their language preference was set to German. Google has to decide whether to prioritize the language of the query itself or serve results based on the user’s preferred language settings. This complicates content visibility because Google is blending query relevance with personalization.
The article we’re discussing goes in-depth on these issues, explaining them for people who may not be familiar with how Google’s language algorithms work.
A new concept for me was the language emotion and country promotion algorithm. I hadn’t heard of it before, but it makes a lot of sense. It explains why some companies create Hreflang tags that misalign languages and target countries incorrectly.
Bill Scully:
Yeah, I see that all the time—companies setting their Hreflang without actually thinking about the languages their users speak.
Bill Hunt:
Exactly. Let’s say 80% of your pages are in French, and you set Hreflang as fr-FR, but 20% of your content is in English. That’s where this algorithm seems to come into play. Gary gave an example of someone launching a .de website that was entirely in English and focused on coding. Since it was a .de domain, Google would recognize it as a German website. However, because the content was in English, it could potentially be demoted based on query relevance.
That’s exactly what we call aspirational Hreflang—where a site’s developers either jump ahead of their localization efforts or only have the ability to define one language-country combination. I reference this in the article and break it down further.
One example was fr-FR, where 80% of the site was properly localized in French, but the remaining 20% was still in English while being tagged as French. Google can detect these inconsistencies.
Bill Scully:
Yeah, and if Google sees that happening at scale, it’s just going to ignore or even penalize those pages.
Bill Hunt:
Exactly. We showed a real case study in the article—one where a company outsourced content creation to a third party that produced over 1,000 pages, all in English. However, their Hreflang tags were set up incorrectly, labeling the content as localized for Greece, the Czech Republic, the Netherlands, Romania, and about 20 other markets. Supposedly, the pages were fully localized, but they weren’t.
In 90% of cases, Google refused to index these pages because they misrepresented their language. The incorrect language settings led Google to flag them as duplicate content, preventing them from being indexed.
When we fixed the Hreflang settings—creating separate XML sitemaps that correctly specified the language—Google started reindexing the content within two weeks. We still targeted Greece, but instead of misleading Google by labeling English content as Greek, we set the language properly. That seemed to align with Google’s algorithm, allowing the pages to be indexed again.
This was another great discussion.
Bill Scully:
Yeah, really good insights.
Bill Hunt:
Before we wrap up, I want to remind everyone that we have a free public Facebook group. Just search for International Web Mastery (Public) or check the link in the show notes. Since it’s a group, you won’t be spamming your friends with notifications. You can ask us any questions, and Bill and I will be there to answer them.
We always appreciate your feedback—it helps us improve future episodes and deliver more value.
In our next episode, we’ll be discussing creative project budgeting. See you then!