As many of you already know, I am Italian and I am a web marketer. These two facts made me discover International SEO very soon because - let's face it - Italy is well-known, but there are not many people in the world who speak or understand Italian.
If you consider the nature of many Italian companies which rely on the foreign market for a good portion of their revenues, you can understand why SEO and International SEO are essentially synonymous for me and many others European SEOs.
This map explains why I must be an International SEO. Image by: https://www.hu.mtu.edu
This explains my interest in how search engines treat the problems associated with multi-country and multi-lingual sites. It also influences my interest in how a company can come in, attack, and conquer a foreign market. I've seen both interests becoming quite common and popular in the SEO industry during these last 12 months.
Many small and medium-sized businesses now have the desire to engage in a globalized market. Their motivations are obviously fueled by expanding their business reach, but are also a consequence of the current economic crisis: if your internal market cannot assure an increase from your previous business volume, the natural gateway is trying to conquer new markets.
My Q&A duties as SEOmoz Associate have made me notice the increased interest in International SEO. Rather than seeing a small number of questions community members publicly and privately ask us, we are seeing many questions based on the confusion people still have about the nature of International SEO and how it really works.
In this post, I will try to answer the two main questions above referencing a survey I conducted a few months ago, which (even though it cannot be called definitive) is representative of the common International SEO practices professionals use.
What kind of solution is best for International SEO?
The answers given in the survey clearly show that people seem to prefer to use country code, top-level domains (ccTlds) against the sub-carpet option and the sub domain. (I'm still wondering what “other” may mean.)
The main reason for this choice is the bigger geo-targeting strength of ccTlds. However, this strength is compromised by the fact that you have to create the authority of that kind of site from the ground up through link building.
Does that mean more companies should go for the sub-carpet option? From an SEO point of view, this could be the best choice because you can count on the domain authority of an established domain, and every link earned in any language version will have positive benefits for the others. Sub-carpet domains could be also the best choice if you are targeting a general language market (i.e.: Spanish) and not a specific country (i.e.: Mexico).
However, there are drawbacks in choosing sub-carpet domains:
- Even though a sub-carpets can be geotargeted in Google Webmaster Tools, they seem to have less geotargeting power than a country code, top-level domain.
- In some countries, users prefer to click on a ccTld than on a sub-folder results page because of the trust that is (unconsciously) given to them.
- If any part of your site is flagged for Panda, the entire domain will be penalized. A poorly organized and maintained multilingual/multicountry site may increase this risk exponentially.
I consider the pro's and con's of both options, and I tend to be strongly influenced by non-strictly SEO needs in my final decision
For instance, it is quite logical that Amazon decided to base its expansion into foreign market using ccTlds. Apart from the SEO benefits, having all the versions of its huge store on one site would have been utterly problematic.
On the other hand, Apple preferred to use the sub-carpet option as its main site does not include the store part, which is in a sub domain (i.e.: store.apple.com/es). Apple chose a purely corporate format for its site, a decision that reflects a precise marketing concept: the site is for showing, amazing, and conquering a customer. Selling is a secondary purpose.
I suggest you to do the same when choosing between a sub-carpet and ccTld domain. Go beyond SEO when you have to choose between these options and understand the business needs of your client and/or company. You will discover there are bigger problems to avoid in doing so.
The local IP issue and the local hosting paranoia
This is a classic issue I see in Q&A. Google personally responded to this issue in an older, but still relevant, post on their Webmaster blog:
"The server location (through the IP address of the server) is frequently near your users. However, some websites use distributed content delivery networks (CDNs) or are hosted in a country with better webserver infrastructure, so we try not to rely on the server location alone."
Nonetheless, in the case that you are using a CDN, examine where its servers are located and check if one or more are in or close to the countries you are targeting. The reason for this examination is not directly related to SEO, but concerns the Page Speed issue. Page Speed is more of a usability problem, but it has an effect on your site's SEO.
Finally, don’t confuse local IP with local hosting as you can use a local IP via proxy from your own server. In certain countries, a hosting solution can be a real pain, and that drives many companies to host their clients sites in servers located in another country.
Takeaway: do not get obsessed by having a local IP for your ccTld site, as it is now a minor factor.
In case you choose the sub-folder option, another important technical aspect is to create separate sitemaps.xml files for every one of them. Again, common sense, but worth mentioning.
The “signals”
If you are going to do International SEO, the first problem you will have is translating the content of your site in the language of the country you are going to target.
It is common sense to use a professional translator (or a translation agency), but some people lack common sense and tend to rely on:
- Automated translation services (especially Google Translate)
- People in-house who happen to know the language the content needs to be translated to
The latter is nonsensical. Professional translators have studied for years and know the nuances of the language they translate, whereas a professional translator will usually translate from a foreign language into their own language, not vice versa. If your translator is bilingual, that's even better.
The first choice is officially deprecated because it is considered (correctly) as a bad quality signal to Google. Even though Google's translator tool was created for this purpose, it seems as if they are sending some mixed messages and I advise you to look elsewhere for translating services.
A professional translation of your content is the best ally for your keyword search.
For example, let's say you want to rank for “car sell” in the Spanish and Latin American market. If you use Google Translate (or Babylon, WordLingo, or Bing Translate), you will have just one of the many variants of that keyword possible all over the Spanish variations map:
- Venta de coches (Spain)
- Venta de carros (Mexico)
- Venta de auto (Argentina)
- And so on…
Even worse, maybe you won’t discover that some countries have peculiarities in dialect expressions instead of “official/standard language” ones, or that people in these countries use both the English wording and the equivalent in their language. For instance, in Italy is very common to say both “voli low cost” and “voli a basso costo,” both meaning “low cost flights."
When I have to optimize a site for a foreign language, I give the translator a detailed list of keywords in which they will:
- Translate properly according to the target
- Use in the translations themselves
Once the site has been translated, I use the Adwords Keyword Tool suggestions copy pasting the translated keywords. The process includes:
- Creating a list of keywords with traffic estimations
- Google suggesting “related to my topic” keywords
- Collecting and analyzing Google Trends information for the keywords
- If you copy and paste the translated page (not just the translated keywords list), Adwords will suggest a larger and more sophisticated list of keywords
- Refinement of the initial list of keywords and, if there are changes to make in the translations due to that keywords analysis, asking the translator to revise them.
Pro tip: Another step I take is to pull the final keywords list into the SEOmoz Keyoword Difficulty Tool to have a complete map of the difficulty and the competitors my site will have to compete with.
Do you think all this is possible using an automatic translator?
A correct translation is one of the most powerful geo-targeting signals a site can have, especially when a language is spoken in more than one country It is an extremely important usability tactic (which is correlated to better conversions), because people tend to trust a vendor who speaks as they speak.
Other classic geo-targeting signals are the use of local currencies, addresses, and phone numbers. Using them is very common sense, but again, some people don’t excel in that field.
However, you may have problems when you target a language all over the world and not a specific country. Obviously you cannot use the local signals described before, because you have the opposite objective. What to do then? I rely on the following steps:
- If the language has variants, try to use an academically standardized translation which is comprehended and accepted in every country that language is spoken.
- You may not have offices in the countries your targeted language is spoken, but you might have a customer care department in that language. Try to “buy” legitimate local phone numbers and redirect them to your real toll-free number, while listing them in your “how to contact us” page on the site.
- Rely more heavily on other signals such as local listings and a balanced link building strategy.
The never ending story of how to implement the rel=”alternate” hreflang=”x” tag
If you reflect wisely about I have written up to this point, be sure to notice that "On Page International SEO" is not all that different from “On Page National SEO”. However, the slight differences arise when we talk about tagging for multilingual and multi-country sites, and there is a lot of confusion about this topic (thanks in part to some contradicting messages Google gave over the last two years).
The geo-targeting tags are needed to avoid showing the incorrect URL in a determined regional Google search. A classic example is seeing a US site outranking a UK site in Google.co.uk, usually due to a stronger page/domain authority. They don’t have any other function than that.
At first sight, its implementation is quite simple:
if Page A (US version) exists also in Page B (Spanish), C (French), and D (German) versions from other countires, no matter if they are in the same domain or different, then on page A you should suggest the last three URLs as the ones Google must show in the SERPs in their respective targeted Googles.
Those tags must be inserted in the <head> section and look like this:
<rel=”alternate” hreflang=”es-ES” href=”https://www.domain.com/es/page-B.html” />
In this line, “es” indicates the language accordingly to the ISO 6391-1 format, and “ES” the country we want to target (in ISO 3166-1 Alpha 2 format).
You can also submit the language version information in your sitemaps. This is the best choice in order to not burden your site code, and it actually works very well, as Peter Handley discusses in this post. Also, they offer pre-existing tools which integrate the rel="alternate" hreflang="x" value in the sitemaps.xml files, as this one by The Media Flow.
Is not so hard, is it? Unfortunately, SEOs had been discouraged by atrocious doubts, especially when their International SEO previewed targeting countries where the same language is spoken.
What is the reason of these doubts? It is the fear for the (substantially) duplicated content those sites (or sub-carpets) tend to have, and the correct use of rel=”canonical”".
For example, in a multilingual site, we have the US American portion of our eCommerce store on www.mysite.com/store/. In www.mysite.com/au/store/ we have the Australian version. Both sell the same products and their product pages are substantially identical.
As we know, duplicated and substantially duplicated content is one of the classic Panda factors. So, does that mean we need to use as canonical of the Australian store pages the American ones? The answer is: no!
Google lists a couple of reasons why this is the case:
- The rel=”canonical” should show a different URL than the one self referential only if the page is an exact duplicated content of the “canonical” one.
- Product pages are not exact duplicates because they have slights differences like currencies, contact phone numbers, addresses, and – if you localized also the text – in the spelling of some words.
In this cases, as Pierre Far wrote in August on G+: "The idea of rel-alternate-hreflang is to help you signal to our algorithms that although these two pages have substantially the same content, the small differences between them are still important. Specifically, the small differences are relevant for users of a specific language (and in a country). By the same logic, the title can be different too."
Therefore, using canonical to direct to a different URL will cause your users to miss a page with potential useful and important information.
What about Bing?
Bing does not use the rel=”alternate” hreflang=”x” markup.
As written by Duane Forrester in this post, Bing relies over a series of signals, the most important being the meta equiv=”content-language” content=”x-X” tag.
Inbound Marketing, Link Building, and International SEO
Now we have our multilingual/multi-country site optimized, but even if we choose the sub-carpet way in order to have a first boost from the existing domain authority of the site and the flow of its link equity, we still must increase the authority and trust of the language/country targeted sub-carpet in order to earn visibility in the SERPs. This need becomes even more urgent if we decided the ccTld option.
So, why is the sum of the budget for all of your International SEO link building campaigns usually smaller than the one of your national market?
Logic should suggest that the first answer (“almost the same…”) was the most common.
The reasons typically used to justify this outcome is that “link building in X is easier” or that “the costs for link building are cheaper." Both justifications are wrong, and here's why:
- To do link building in every country is harder than it seems. Take Italy, for instance. is not so easy as you can imagine. In Italy, the concept of guest blogging is still quite “avant-guarde.”
- To do #RCS which will earn your site links, social shares and brand visibility is not cheap. In Italy (I'm using my home country as an example, again), the average price for a good infographic (not interactive nor video) ranges between $1,000-1,200 US dollars.
My suggestion is to investigate the real costs for International SEO content marketing, link building, and – eventually – social media marketing. You should also ask the opinion of local link builders if you can, even in the common case that you will perform the link building campaigns internally.
In fact, those local link builders are the best source to explain what the reality of link building looks like in their countries. For instance, how many of you know that Chinese webmasters tend to prefer a formal outreach contact via email than any other option? I didn't know until I asked.
Modern-day link building does not mean anymore directory submissions, dofollow comments, or forum signatures than it used to, but it has evolved into a more sophisticated art which uses content marketing its fuel and social media as its strongest ally.
Therefore, once you've localized the content of your site accordingly to the culture of the targeted country, you must also localize the content marketing actions you have planned to promote your site with.
Luckily, many SEOs are aware of this need:
And they usually work with local experts:
If we consider SEO as part of a bigger Inbound Marketing strategy, then we have to consider the importance Social Media has on our SEO campaigns (just remember the several correlation studies about social signals and rankings). So, if you are doing International SEO, especially in very competitive niches, you must resign yourself to the idea that you will need a supporting International Social Media strategy.
Conclusions
International SEO for Google and Bing is SEO, no questions asked. It is also not substantially different than the SEO you do for your national market site.
Sure, it has some technicalities, but you may need to use them if you want to target other languages spoken in your own country, as Spanish is in the USA. All the rest of your SEO strategy is identical, other than the language used.
All the concepts related to Link Building and Inbound Marketing in International SEO and SEO are the same. The only difference lies in what tactics and what kind of content marketing actions works best from country to country.
What can really make International SEO much more difficult than “classic SEO” is one basic thing: not understanding the culture of the country and people you want to target. Don't make the mistake of having your International sites written by someone like Salvatore, the literally multilingual character of "The Name of Rose" by Umberto Eco.
Ron Pelman in the role of Salvatore in "The Name of Rose" (1986) - Twentieth Century Fox Film Corporation
Bonus
Here are a few other posts you will find useful about International SEO:
- Webmaster Central office hours about multinational and multilingual sites, Google Hangouts organized by Pierre Far
- SEO for Multilanguage Multicountry Projects, a slideshare by Ani Lopez
- International and Multilingual sites: The Criteria to Establish an SEO Friendly Structure by Aleyda Solis
Hi Gianluca,
First of all great post, if you don't mind i would like to add a few points -
Regarding Alternate Tags -
The biggest challenges are faced by Multinational companies that sell thousands of products and have different country specific pages that have the same content. In such cases adding multiple alternate tags becomes a huge issue. For example if I have an electronic product page and that same product is being sold across multiple countries that have different country specific URLs and country specific content, then the webmaster will have to add multiple lines of Alternate Tag code on each product page. Imagine the plight of the webmaster in cases where there are 1000 of products across 50 - 60 countries.
However one seriously cannot get away with this issue. In the Mid April and May of 2012, where Penguin and Panda were on a vengeance, I noticed that a lot of websites fell in rankings, in fact in such cases Google started showing those websites which it considered as the original source. For example, in UK Google started ranking US website despite the fact that there existed a UK website. The theory here being that, due to the same content i.e. US and UK websites were selling the same product which had the same specifications (content), Google started ranking that version of the website which it thought was the original source of content i.e. the US website. Why did Google do that? The reasons can be many, higher domain authority, better link profile etc etc. But what intrigued me was the fact that it did not remove the domain from the SERP. With the Panda update that deals with content scrapping/duplicate content, emerged a problem that I call "Geographic Bundle of Confusion". The problem did not persist only in UK or US, same was the case with India where US website started ranking in place of the India version of the website. In South America, there were so many cases where Spain version of the website started ranking in place of the Argentina version of the website, simply because both these websites had the exact same content and Google did not consider the latter to be original source of content.
In such cases the significance of Alternate tags cannot be stressed enough. But one might argue, what is the harm, as such the domain is ranking, how does it matter if it is the US domain or the UK domain? The answer is measurability. Also one should not ignore the fact that every Geo specific website is always designed keeping in mind the local needs of the people and users.
From Geo perspective, there are a lot of things that the webmaster/SEO can do to help the website -
Hope my comments were useful.
Cheers,
Sajeet
Hi Sajeet, thanks for your deep comment and for being here.
The sitemaps.xml solution is the one I prefer and suggest (also in the post) for cases like the one you cited.
For PDF files and any other not HTML document, you can use the rel alternate in the http header, as suggested by Google itself:
HTTP header. If you publish non-HTML files (like PDFs), you can use an HTTP header to indicate a different language version of a URL:
Link: ; rel="alternate"; hreflang="es"
About the geo tags... honestly I don't agree, as I don't think Google really take them into account (as it happens with the meta equiv="content-language" content="x-X" tag).
Finally, to geotarget the domain/subdomain/subfolder via GWT it is indeed an obligated task for a correct International SEO (but only if you are targeting country markets, not languages all over the world)
You can take care of PDFs via the Sitemaps option, too.
Gianluca, you made my day!)) Absolutely awesome post.
I was looking info about this issue coz I'm working not only locally (from Ukraine) but also promoting few sites in the USA region.
So what can I admit, that your second image with graph under the "What kind of solution is best for International SEO" is absolutely true. I just noticed this from personal experience... Great job.
Thanks!
When thinking about how writing the post, I finally decided to follow a chronological working process but using somehow a Q&A style, supported by the result of the survey few months ago and still haven't shared.
About the "best solution" for International SEO, I confirm what I wrote in the post. The best solution is usually not determined by your SEO needs, but by the company general (and more important) business objectives.
What do you think about not only supplied argumentsnot only but effect of lang= tag on donors pages? I couldn't figure effectiveness of that thing yet...
Mmm... may I ask you to rephrase the question, as I am not sure I understood it well. Sorry... I'm Italian :)
I mean <html lang="en-US" xml:lang="en-US">
Honestly I don't have an answer, sorry. Haven't really cared about that markup.Maybe the best thing to do should be experimenting it and see what the test results are.It could be nice YouMoz post :)
Thanks a lot for sharing that make my day also ;) The advise not to add canonical to alternative versions is simply awesome as I was looking for the info since long time and followed a thread where +JohnMueller has also answered but was vague.
Thanks, but actually I simply put in evidence what Google itself finally said about that issue. The problem is that it said so many things that, if you don't follow International SEO constantly, you may have been behind in the information frenzy.
Thanks for sharing that link, Mediative - up until now most of the stuff I'd read on the 'canonical vs. hreflang' was simply: don't do it! Interesting to get more of an insight into it via that forum post.
This is a great post Gianluca, published at the right moment!
Reading about the duplicate content issues and the Panda risks I would like to ask you:
1. Do you think sites on different ccTLDs with content in the same language run the risk to get hit by Panda? For instance, a US established site does exist for some years and 2 new sites get launched on different ccTLDs (e.g. .co.uk, com.au). Assuming all these sites are very similar in terms of content , would all 3 sites run the same risk to get hit by Panda or just the new ones?
2. Do you think that implementing the hreflang on the above 3 sites would eliminate the Panda threat? If not, what could be done other than writing unique copy?
3. In your experience, what is the percentage of unique copy each site should present to be confident that there won't be any duplicate content issues implications like Panda?
Nice questions.
Let me start from the last one: no, I have no formula about what percentage of unique copy each site should be present to confidentally think it won't be hit by Panda.
It may sound "obvious", but the only correct answer is: "the much % you can".
For instance, instead of doing a crazy-scientist job of calculating percentages, we should think about what can differentiate a product page from the same in the others sites, while adding values to the users at the same time. An idea should be to add user generated content, as reviews of the products.
About the implementation of hreflang also as a way to avoid Panda due to the obligated similarity same language pages but in different country targeting domain names or subfolders, if you read again what Pierre Far said (cited in the post), then we should be confident that the hreflang tag have also this function, because it is alerting Google no only that those pages, even if are substantially identical, nonentheless they have slight differences that makes them unique and useful for their targets.
By the way - as said answering to question 3 before - if you are able to differentiate those very similar pages the better.
And related to your first question. I think that is quite probable if you don't use the hreflang markup.
Thanks for the detailed reply.
It is very difficult to differentiate the content appearing on different ccTLDs when the business offering is a service rather than some products. These services are always the same and do not have anything in particular to make them unique in each country.
So, from what you said it sounds like the only way forward is rewriting the copy.
Excellent post my friend, and extremely timely as I'm starting to think about the SEO plans forward for our various international regions.
Rather than talking about "translation," which I think can easily lead to the shortcuts like Google Translate and Babelfish approaches, it's better to think of it as interpretation and localization. Consider the United Nations, which doesn't rely on translators, but interpreters, who must not only take in the original words, but those meanings, and then interpret them into the other language in a way that retains the meaning. Obviously this is considerably more challenging and involved.
Have you experimented at all with the "ccLinks" technique that McAnerin talks about, using XML sitemaps and 302 redirects on the ccTLD for a .com version? The article is a little dated, so I thought I might ping Ian before putting too much consideration into it. I'd ideally prefer to go straight ccTLD's, but as you said, there may be other non-SEO -- gasp! -- reasons that may require retaining the .com/directory/ approach.
I have a feeling I'll be revisiting this post and the links quite a bit over the next few months!
Grazie
Grazie a te Brian :)
Thanks especially for your suggestion of using the word "Interpreter" and not "Translator". It is indeed more exact and helps understanding how really you have to treat content in International SEO.
About the ccLinks practice, no, I haven't use it, also because - even if it may be effective - it seemed to me a little too on the edge, as 302 are temporary, and you cannot be eternally temporary. But, maybe I will experiment with it.
An absolutely fantastic post, Gianluca! Face-snappingly good! ;-)
Seriously though, a very timely post for me as well, as I'm juggling with hreflang for a big client. Their situation is more complicated than your average site. They're a financial company and by law they have to show two different versions of the site within a country to two different types of users, so in domain.com/uk there'll be two close duplicates as well as within domain.com/us, etc. I pitched some questions to Pierre very recently, which might be useful to other 'hreflangers'. What was interesting is that Pierre seems to suggest that canonical can be used within a set, so all pages within domain.com/uk can refer the hreflang alternatives as well as canonicalise a version within itself, within each country. Does that make sense? Is that even possible? Can anyone verify? I can give a clearer example if needed (I've given a small one in that link above).
Good on you for sharing Pete Handley's posts - I was going to do it myself if you hadn't have done (as they're really useful) but he deserves a dofollow in-post link more than an in-comment nofollow link, that's for sure! ;-) This post gives some good insights, too. There seems to be a severe lack of hreflang case study posts - I'm hoping to write one once we've implemented it for this client (in the next month or two), as they think they've been heavily Pandadised. But I implore other SEOs dealing with hreflang/int. SEO to share their stories and successes, as it's still rather a turbulent area by the sounds of it, given some people's experiences.
Hi Gianluca - thumbs up as always ...
I wanted to add the following information: Sub-carpets can only be geotargeted in GWT if it is a gTLD - so this has to be considered in dealing with International SEO, too. All the Geotargetable domains can be seen here: https://support.google.com/webmasters/bin/answer.py?hl=en&answer=1347922
Thanks Petra,I think that Google decided so because to target a subfolder in a ccTld should be somehow contraddicting somehow the same nature of a country code top level domain, which is meant for targeting a specific geographic market.
So you're saying that if you have a .com that's split into .com/uk, .com/us, etc. that you can still geotarget different parts of the site using GWT? Didn't know that. Good to know. Thanks Petra!
I'd heard that you could only geotarget different domains - sounds like I was told false info...
I'm assuming doing both hreflang and geotargeting is better than just doing hreflang, or are there risks in doing both (e.g. potentially confusing Google)?
Hi Steve,
if you use a .com Domain you can set up e.g. .com/uk to Great Britain, .com/us to the United States, .com/es to Spain and so on. Of course each with an individual setup in the GWT.
But what I wanted to point out is, that with a ccTLD like e.g. .at I am not able to do that (this is of course already setted to that country which it presents).
Most of our clients do prefer a ccTLD so this is an important consideration before a domain is set up regarding to International SEO. If possible we work with geotargeting and hreflang (additionally to seperates XML sitemaps per language, local seo etc.)
Great post Gianluca, you have given brilliant content about the international SEO, there are many marketers who have issues about the multilingual websites and multinational websites you have given really good solution for them.
Hi Gianluca, excellent post! I don't do International SEO myself, but this is one of the most thought out posts I've seen on the topic. Aleyda has provided some great information too!
I am certain many people will use this post as a reference for their own work with sites in different languages/locations. I know many of my own SEO searches are beginning to include "seomoz" as a parameter.
I think your comment in the last paragraph is very true. Just because you can translate your content into another language doesn't mean you can send the same message. Culture matters everywhere so make sure you understand the people and not just the language when you move into international markets. Great point!
Great Post Gianluca,
im facing myself lots of this issues with international SEO and i have to say that is a real pain. Even with this information it gets really complicated to do SEO in another language, even more if you dont know a bit about it. (german for example is driving me crazy)
In fact, u hit my biggest problem with all this: fear!!
I´ll post a comment soon with my experience with href lang, subfolders in webmaster tools, since im not very happy with the results obtained from hreflang
As i peviously said, great post and really good ideas to use local phones! Coud you give me an example of that kind of service?
Best!
For local phones... Skype offers you the possibility to buy (or better, rent) a local phone number in a great number of countries.For instance, let's say I have many clients in the States who don't use Skype. If they need to call me they should need to make an Intercontinental phone call. But if I rent from Skype a local US phone number, they can call me on Skype with a local call price.
Great post, as i said on twitter.
I would like to share some horror stories: i've worked with some international agencies on Seo projects regarding italian customers and except for one case, it has been a bloodbath. I wish i could do names without being sued, but i can't.
I have personally experienced:
english based keyword research translated using google transaltorMy Seo copywriting validated through the use of google translatorLink building machine-made for italian keywords using spinned english texts (hello penguin!)And those scumbags seo agencies report millions of dollars of revenues each year :-/
Hi Gianluca, a fine piece indeed that is also close to my heart.
It's definitely a challenge to balance inbound marketing for an international site and it's often an underestimated cost of being fully international.
The issue of scale really comes to the fore when you also factor the possibility of multiple language or local social support for the brand/company. eg have one (twitter) brand account per territory in each language or a central one. One Google Plus company page with Circles per territory or separate for each territory? Planning and ambition has to be very clear for international SEO. I do wonder how these interesting survey results have evolved over time.
ps . Please don't snap my face when I remark that you have, possibly unconsciously, used the Spanish word for folder: "carpeta" as in subfolder, throughout the post as sub-carpets, as in the upholstery meaning:)
LOL... it's you who I have to snap the face of, but myself... which is something I don't know how can be done, honestly :)
And you are quite right in saying that there are many questions about the strategic plan, which must be answered before we jump into the international market.
Those are the real difficulties of the International SEO in my opinion. Unfortunately, the vast majority of companies do not realize it until too late.
I figured that sub-carpet was the same as sub-folder or sub-category. And since that is the case, can you confirm this?
"Even though a sub-carpets can be geotargeted in Google Webmaster Tools, they seem to have less geotargeting power than a country code, top-level domain."
I don't think that you can geo-target by sub-category/folder in Webmaster Tools. Only by subdomain. At least that's my experience. When I go to Google Webmaster Tools under "Configuration --> Settings --> Geographic Target" that will geotarget the entire subdomain.
Is there another way you're aware of to geotarget by sub-folder? If so, I'm very interested in that. We are internationalizing many sites with 100's of thousands of pages into over 60 countries. Needless to say this post confirms much of our strategy and comes very timely for us! Thanks!
I am glad you were able to understood that i was meaning subfolders when writing sub carpet :).
Answering to your question, actually you can geotarget any subfolder in GWT.Simply you must add the subfolder as it was a new site.Then, you proceed as usual.Be aware that this works only if you are using a generic domain name termination, not a top code country level one.
ahh, yes, that makes sense. I wasn't aware you could add a verified site that way. Guess I never tried it. We probably won't need to do it that way though because all of our generic tld's (.org, .com, etc) are global sites and not geotargeted content. All of our geotargeted sites are on cc tld's or subdomains that we can geotarget in Webmaster Tools.
But glad to learn one more thing about GWT that I didn't know!
gfiorelli1 , love the xhreflang part which i was working very hard lately for a client. TO add to that , when using xhreflang tag in your head your dont need a rel canonical to indicate the preffered version of the page to search engine.
So as an exemple , if i target some content in english , and i have a en-us ( USA ) version and a en-gb ( UK ) there no reason to add the canonical to it even if the content is very similar except culture information ( as currency , time of the day , and stuff like that )
Hope that help !
Best,
Yan
Gianluca,
Once again you get a Wow! from me. I am impressed by the this well imagined and well constructed post and I only regret it has taken a few days to get to it. Two things stand out, first the excellent discussion around rel="canonical" and rel="alternate" hreflang="". the other is around the nuance you present like the Chinese preferring direct email contact. Well done. Your post should be a primer for anyone seeking to expand into International SEO.
A question that arises is with a client who expands into truly disparate languages/characters like German/Chinese/Arabic or a language with Cyrllic characters,etc. Do you believe if they are already getting traffic to their English speaking site from these disparate countries and now wish to expand into them, that the weight of their current domain is more value than having separate ccTLD's (say at DA of 40) and in a cost benefit analysis, is it more cost effective to create a single ccTLD site for say Chinese if you are an American firm or to add to the current site and make all the coding changes that will be necessary?
Thanks again for presenting actionable data in a well thought out format. All the best,
Robert
Edit to add a question: If you are getting good traffic to your American/English only site and converting it, is this a case where having a site in the other country that is a ccTLD would actually be an enhancement due to new traffic or would the same be true by having the alternate language/country on the current site?My concern being that once there was an alternative, you would lose some of what you had gained in traffic and eventually ranking to the newer site.
HI and thanks.
Answering your question:
first of all I would not put Russian in a subfolder, especially because subfolders don't perform well in Yandex, which gives a huge competitive advantage to the .ru domain names sites.The same it's valid for a chinese version of the site.All the others, eventually, could be created in subfolders.
Related to the traffic... nice question. In general it would not decrease; instead it should increase, because using the languages spoken in those country you are targeting and correctly implementing the hreflang markup, actually you are targeting better those markets that now are part of your original site's traffic.In the specific of your main site, yes, it is quite probable that you will see a decrease in traffic from Google coming from the countries you have created new sites (not valid for the subfolder case).
Sorry to post this separately but another question that's come up since: what happens if you implement hreflang and Bing's alternative (meta equiv="content-language" content="x-X" tag) together?
In theory it sounds like it should be fine, but has anyone had any trouble in doing so? Any case studies/success stories? Anything we should know about?
very interesting post Gianluca!Considering my experience in international Seo (as I am italian too) I would say there is not a one fits all solution. There are many factors we need to consider usually related to the specific markets we work in. I agree there is a global trend but I believe it is mostly produced by the amount of english-oriented sites. Structure of the domain, Hosting solution, language management are topics that only big global sites really started working on with a seo approach
Great Post! This post clarified a lot of questions I have regarding to setting up an International Website. However, I have one question that remains unanswered and I've also asked this question in the comment section of Rand's International SEO white board friday.
My company is planning to purchase a .hk ccTLD and create blank services pages for each service we offer in HK. Then we are planning to redirect each .hk service page including the home page back to our .com website and .com service pages.
The question is, do we still get the benefit of creating a new website with .hk ccTLD in the search engine and will Google still see this as targeting people in HK.
Please advise.
Thank you!
Great post!
I need some help on this subject??
I'll use france(/fr) as a location example.
I have a site hosted .co.uk domain www.stevencoogan.co.uk and have just install a translator which kindly puts a /fr for france infront of the top level domain
Is it better practise / more effective to use a .com domain? so stevencoogan.com/fr for glocal seo?
Thanks, Steven
Hello,
I find the subfolder route the easier way but what is better for a european ecommerce that can afford to manage only four languages?
Subfolder for each langauges:
site.com/en (english content)
site.com/it (italian content)
site.com/de (german content)
site.com/fr (french content)
or
Subfolder for each EU country but only such four languages will have unique content the other will have all the same english content. Example like yoox website.
site.com/uk (english content)
site.com/it (italian content)
site.com/de (german content)
site.com/at (german content)
site.com/nl (english content)
site.com/pl (english content)
site.com/dk (english content)
site.com/se (english content)
site.com/cz (english content)
..
..
..
etc.. for each EU country with the same english content...
Thanks for your attention
Great post Gianluca!
What are you views about IP sniffing to deliver local content to different users? Have you ever tried this in any international SEO strategy?
I don't use it for the simple reason it usually developed badly, causing people to see under www.domain.com their language version... causing serious SEO issues (i.e.: links for a language/country version pointing to the domain name, which should be related to another language/country).And it is also a practice Google suggests not to use, also for usability. Infact, not necessarily one person going to your site wants to see only his country version of the site itself.
I'm trying to convince a client not to do it but they seem stuck with that idea. I couldn't find any article from Google saying that they don't suggest using this option. Do you have any in mind?
They don't want to use subdomains, or even folders...we'll see how they set it up but I expect a mess on this project :)
Thanks for your answer.
"Avoid automatic redirection based on the user’s perceived language. These redirections could prevent users (and search engines) from viewing all the versions of your site." - support.google.com source
Meh. That's just general advice.
Ask your client this: is it absolutely necessary to deliver the Irish user exclusively to the Irish site? Should the Irish user be able to see other sites?
If the answers are yes and no respectively, you can get away with IP sniffing & server side redirects as long as you do not force redirection on American IPs (Note: Gbot still crawls from the US, but if they change, this won't work). The downsides; usability issues, and the fact that American IPs can see anything they want.
You could also use client side redirection (javascript) with even American IPs with no real consequence (Gbot shouldn't* execute the JS). This is a bit more flimsy, and won't work at all with JS disabled users. Maybe use https://www.wipmania.com/en/api/ for example, and maybe test to see what pisses people off less?
FYI - This doesn't mean that I recommend it! ...and you'll still need somewhere to host the region specific content.
Thanks Gianluca for this great post from a fellow Italian :)
In the situation of an international company having a global .com site and a US-related .com site, with the global winning over the US for US-based searches,
I was wondering what would is best to set in WMT for the global .com site as geographic target - "unlisted" or not setting the target at all?
I had no big luck with either so far but wanted to understand if there was an opinion about the two options.
Ciao!
Antonio
Good question Antonio,
actually:I would not geotarget the fist .com (the global site);I would target the US for the second .com (the american one)I would use hreflang on both, in order to be sure that Google shows the American .com site for searches done in the US, but the global .com in any other part of the world.Actually I cannot tell if that could work, but that's what I'd do. Try it.
But, yes, if in GWT we could exclude countries from being targeted and not only targeting one, that would be of great help.
I have a website which deals with online gifts and my target audience is in UK, Ireland and Canada. I didnt wanted to go for cctlds for all three countries. Instead of that, I am using my .com website to target all three locations. I have created sub-directory for every country in this way: www.mydomain.com/uk for UK, www.mydomain.com/ireland for Ireland. Further to this, I have verified each of these three sites in webmasters separately in order to set the target location. Also, i have varifid my www.domain.com in the same account. Now here is my question:
I have to create sitemap for each subdirectory, I use https://www.xml-sitemaps.com to generate sitemaps. But problem with this tool is that it generates sitemaps as a whole as opposed to sub-directory level. I tried using www.domain.com/uk as url, but it always includes UK, Ireland and Canada in the resulting sitemap. So, I then decided that I will remove every instance of other sub-directory url from the sitemap and will be doing same for all three sub-directories. If anybody knows a better way, please let me know. Once I have three separate sitemaps for three subdirectories, i submitted them in their respective google webmaster profiles under sitemap sections. Apart from this, as i verified my root domain in the same account, I created a sitemap index file and them submitted the file in the root level webmaster account. My sitemap index file looks like below:
<?xml version="1.0" encoding="UTF-8"?>
<sitemapindex xmlns="https://www.sitemaps.org/schemas/sitemap/0.9">
<sitemap>
<loc>https://www.domain.com/uk/uk-sitemap.xml</loc>
<lastmod>2004-10-01T18:23:17+00:00</lastmod>
</sitemap>
<sitemap>
<loc>https://www.domain.com/ireland/ireland-sitemap.xml</loc>
<lastmod>2005-01-01</lastmod>
</sitemap>
<sitemap>
<loc>https://www.domain.com/canada/canada-sitemap.xml</loc>
<lastmod>2005-01-01</lastmod>
</sitemap>
</sitemapindex>
Directory level sitemaps will handle per directory urls and indexation and root level sitemap index file will make sure that all sitemaps are their to index. You can also use https://www.themediaflow.com/resources/tools/href-lang-tool/ to generate the sitemaps.
Great post Gianluca!
What do you think about the benefits of local hosting? I mean, do you think that is much better due the download speed? I know that can be very expensive global strategy but if you want to hit a market in special is a good aproach.
So in you rel alternate example, would you have 4 lines of code on every single page? So 16 lines of rel alternate code in all?
Don't consider this comment of mine, as it is totally incorrect. I don't cancel it because it is a good example of not writing when you should be sleeping :). Fatigue made me totally forget the basics of hreflang best practices (and if this happen to me, imagine...)
No, they will be 12, because you have to show the alternative URLs to for the given languages and countries, hence you don't need to show as alternate the same URL the markups are.
And exactly because can happen that a multilingual/multicountry site targets many languages/country, in order to avoid to burden the code with a long list of rel="alternate" hreflang, Google developed the option to implement them in the sitemaps.xml files.
Hi Gianluca, maybe I've misunderstood, but are you sure it's not 16? Doesn't each URL have to reference itself as well?
"If you have multiple language versions of a URL, each language page in the set must use rel="alternate" hreflang="x" to identify all language versions including itself" (source - emphasis added).
No no... you and Redgatst are totally right, and it was me totally confused while reading the comment... late hours was having a disastrous effect :)
Yes they are 16. A reason more for going the sitemaps way.
Using this example, my US site (https://www.mysite.us/page-a.html) would include the following tag:
<rel=”alternate” hreflang=”en-GB” href=”https://www.mysite.co.uk/page-a.html” />
And on my UK site (https://www.mysite.co.uk/page-a.html), it would include:
<rel=”alternate” hreflang=”en-US” href=”https://www.mysite.us/page-a.html” />
The above example I can use for my US/ UK and DE/ AT sites but before I confuse myself any further, how would a global hreflang tag look if targeting English language but no specific country?
how would a global hreflang tag look if targeting English language but no specific country?
In that case you would use only the language ISO code ( "en" and not "en-US")
My initial thoughts was to use en-EU but your suggestion makes more sense. Thanks again Gianluca and happy weekend
Gianluca's right, but I just wanted to add that I don't think there is an "en-EU" tag. I think it has to be a specific country - I believe codes for multiple countries and/or continents (such as EU) will not work. As Gianluca says, just the language for Global/International/Rest of World areas of the site.
Thanks, appreciate your comment.
Excellent post and impeccable timing as I have just recently started working with a global client. Interestingly enough, I created points myself as I had planned to document the progress and submit the results to SEOmoz, maybe I still will but this post does most of the job.
Here is the approach (raw notes) I am using. Very similar to what you mentioned:
If there is anything you feel I am missing, please do comment. Thanks!
Substantially your checklist is right, but I'd pay a lot of attention to the use of rel="canonical" in same language but different country targeting sites or subfolders (check what written in the hreflang section of the post).Give a lock also to the link to the Aleyda post linked in the Bonus section of this mine.
Thanks, I missed the bonus section.
If I have sites targeted to the US, UK and a global site (all English):
www.mysite.com
www.mysite.co.uk
www.mysite.us
Would I need to implement rel=canonical on each site?
I want to rank in local search engines, but figure the href-lang tag would supplement that.
Only difference in sites are currency/ local spellings. URL structure is pretty much the same on all sites.
I tend to suggest always the use of the rel="canonical", because it is the best way to deal with classic internal duplication problems.But, in your case, even if the sites have some pages which are substantially identical, I won't use the cross domain canonical version, in order not seeing my UK pages, for instance, disappearing in Google.co.uk as being substituted by the "canonical".
Thanks for clarifying.
I was 85% sure but now I'm even moreso from your comment.
Thanks again
A lot of great stuff here. We're going to be discussing the topic on this week's #SEOpub Twitter chat with Kristjan Hauksson. It would be great if you could stop by and join the discussion as well.
It would be a pleasure, even though I'd need to check the time zone :)
Thank you for this post. I've been reading the bookGlobal Search Marketing, which is more about understanding the language changes and differences in how we speak. As well as you mentioned how in China they prefer an email, another thing to figure out, something as simple as how people connect in different countries.
It's certainly a huge endeavour if you're trying to market in many countries to understand all of these differences before you even start your actual site SEO, but like you said just as important.
Glad you cleared up a lot of technical questions that I had about it, very well done post!
I've just got a quick question that I'm hoping someone will help me with.
It might sound like a simple question, but.... I've noticed that the example given has have the hreflang as ="es-ES". Does the second code have to be in capitals?
in the one i'm working on i have them as "en-gb" and "en-us" and I'm doing the sitemap option. Only implemented it two days ago so its early enough to change, I just want to make sure having the "gb" and "us" in lower case will still work?
Actually that second code is the country code accordingly to ISO 3166-1 alpha 2 standard, so, to be respectful with it you should use the capital letters.And, actually, it is also better for distinguish language from country in cases where the same code can represent also the language, like "es-ES" (spanish-Spain). Said that, Search Engines are smart enough to understand it also in minuscule.
Hi Gianluca, Im an SEO and I actually do SEO. Im italian like you, and I can confirm what you wrote in your post. When you state that doing International SEO is not different substantially, you are right, but as you already know, the language is a way of thinking, so Americans, speak, and THINK (especially) different. So, when people think in a different way, speak and SEARCH different. For example, Italian language have many words to say and explaine a concept, so in our daily work we have many keywords to work. I think, personally, that we can work much than english on the long tail.Searching, is a daughter's of thinking,
Of course, you are absolutely right when you affirm that you can't hire a person like Salvatore to do International work..
tks for posting your idea.
Ciao Franco,
you are right when you talk about how different people do a search from country to country (hence, culture).
But, if you think it, that is something you must already do if you are working for one country alone. In fact, in the States people search differently depending on their culture, ethnicity and even their education level, because in English too you can say the same thing using different words due to its same history as a language:"noble", literary words coming from a Latin root (i.e.: street)popular words coming from a Germanic root (i.e.: road)When you have to do SEO, regardless of the country, you must first understand and know how people search... this analysis, then, open you perspectives and opportunities as the long tail one you described for the Italian language.
Absolutely Giancluca, right insight!
In one State, can live many cultures, so many ways of typing queries for saying the same thing, so discover other opportunities.
And about the history of language, you are absolutely right. Thanks a lot
Ciao
Hi, great post. I have to agree with you on the fact that the best signal is the correct language. One of my proofs is one small adsense site which I own, and is going very well in Google.gr, the domain extension of this site is { .in }. The hard is to build local backlinks, which is very hard in the Greek market, (they dont link if they dont know you 80% and rest 20%need content) .But I saw good results from english sites when I link to this small sites. But I saw even better from a contextual link which is coming from a .gr domain.
Gianlucca this post is excellent!
One of the most important pieces of this is the budgetary issue. Companies usually look at foreign language/country initiatives as experiments that have to prove themselves before earning a larger budget, but in order to do so, they need to invest more! It becomes a chicken vs. egg situation.
Because we have encountered this issue many times, we now optimize for a shorter list of words that we would for someone who comes to us with an existing Spanish language site, knowing that we will have to prove ourselves for that list in order to get more resources.
The inbound marketing aspect of this is also a huge factor. We now suggest that companies have separate social media profiles for each language (in the case of SMB's) and each country for enterprise level. It means starting over from scratch, but the end result is higher engagement and a better customer/company relationship overall.
I must admit, this is the first time I've seen the term ' sub-carpet domain' used and wasn't 100% sure what it was referencing. Apart from that, great post, the first time I had to implement rel=”alternate” hreflang=”x” it took me a while to get my head around it. However, I wasn't aware that you could add these tags into the sitemap! Added to to-do list, thanks.
Thanks :).And sorry for the confusion... sub-carpet instead of subfolder was a totally unconscious use of Spanish into English by my side. Check more why in this comment thread if you want to have fun.
I was lost on International SEO , this is a great post with some great insight and data to work with. rel=”alternate” hreflang=”x” is relatively new to me and certainly looks interesting. Thansk for the post gfiorelli1
Thank you so much for this. I was just starting to get interested in international SEO because I think eventually it will be relevant to our company. Thanks for the information at least now when I tell people to get a real translator I have something to back it up with.
Bongiorno Gianluca
One thing that i'm very interested in is how Google's Algorithm treats domains owned by foreigners.
In my opinion, its being developed based on the U.S market and U.S searchers.
I might be wrong about this but there doesn't seem to be much discussion about it.
It would be interesting to know how many links are given to non U.S owned sites compared to U.S owned sites. And the social figures for shares, likes and follows relative to a countries population and/or online presence.
If Google was owned by the Italians, don't you think the algorithm would favor Italian owned sites?
Its like Baidu. This search engine seems to favor Chinese owned sites over foreign owned sites.
Most of my traffic comes from the U.S which means America is a great source of business and without the U.S market it would be a different story. Can't say the same for Baidu in China and not sure if translating a site into Chinese would be worth the effort.
Why would any country allow foreigners a fair chance to take away online business especially if they don't live in that country?
Mmm... I am not sure if I am following you, but - IMHO - to think that Google is giving more power to domains owned by US owners. If you do a simple test in Google.it, you can easily see with an Whois check that in great majority of the cases the site owners of the sites ranking in the first page of the SERPs are Italians.
The fact is that Google.com is at the same time the search engine for all the world but the search engine you have to work on if you want to target the United States market... a Google.us doesn't exist... so, just because of that you may see more "US owners" site winning the ranking game there.
About Baidu... if you are failing with it, it is not because you are not Chinese, but because - in this case your theory is partly correct - it has been created thinking to the Chinese searchers (also for obvious political reasons). So it is failing because it is not in Chinese (as you said), it is not, maybe, on a .cn domain name (clearly favorited by Baidu, as the .ru one by Yandex in Russia) and because, and here it comes the culture, it is not designed for the Chinese public.
Having worked on international sites before the element which resonates with me is using internal staff who speak the language you're optimising for. Speaking the lingo and optimising for it is two different things, especially if keyword research is concerned. Spend the money, get a professional involved and get it right first time.
International SEO is a really interesting subject, there's so much opportunity in it yet not too many people have really understood what to do, mostly they just simply translate their content and redirect their users