How to Translate Your Content Without Hurting Your SEO

There are a lot of websites out there in the world, but it's easy to forget that there's an entire universe of digital content you probably never see. 

How many websites do you have no reason to visit, because you can't read what's on them regardless? How much content is out there in Spanish, or Japanese, or Arabic, or Russian? The answer is: a lot.

As a web user, I sometimes end up on one of these sites and rely on machine translations to read the content. But, occasionally, one of these sites has bilingual content. Instead of relying on me to figure out translations that could be wrong, they provide their translations, presumably with a greater degree of accuracy.

As a website owner, I (and my clients) need to decide whether or not I want to translate content. And, if so, how to do it successfully.

Why and When Translating Your Website is a Good Idea

So, first: when should you consider translating your site, and why?

The biggest reason is just that you have an audience that spans multiple languages, and you want to better serve them.

I see this a lot with local business websites in areas where there's a significant ethnic minority population. Many places along the southern border have a good use for Spanish-language content, for example. Various cities around the country can make good use of various languages, depending on their local community makeup.

It's also useful if you're a business that can ship internationally, even if you don't do much international business yet. You might have a lot of website visitors from Brazil or parts of Europe or what have you. Why not try to capture that traffic by reaching them in their native language?

Research has shown that most people prefer buying from websites written in their native language, which sounds pretty obvious when you think about it. Navigating ecommerce is tricky enough with sites that aren't the big names like Amazon, and adding a language barrier makes it even harder. A person has to be very motivated to try to buy something through a translator, you know?

It's also important for clarity. Yes, anyone with a web browser can translate content, either automatically, using MTL, or using AI these days. But we all know how terrible those translations can be. Providing your own translations allows you to validate what they say, so you guarantee the information you provide is accurate.

Of course, all of this feeds back into commerce. Informational websites can reach more people with more accurate multilingual content, and make more money from display ads. Ecommerce stores can reach larger audiences and sell more products. Translation expands your horizons and broadens your ability to sell.

The Wrong Way to Translate Your Content

All of this is great, right up until you sit down to translate your content and realize you have no idea how to do it properly.

Worse, if you do it wrong, you're going to tank your SEO, and that can take years to recover in the worst-case scenarios. 

So, what are the major mistakes to avoid? Here are the things I've seen wreck a website and make translation a waste of effort at best or a devastating problem at worst.

Trusting raw output from a translation app.

Language is very difficult to handle even by people. It takes years to learn a language and even longer to master it. It's even more effort when you're learning a language in adulthood. And all of that is before you even get into dialects, regional variances, and other elements of localization.

For most of digital history, computer translation has been subpar to say the least. Most languages have a lot of words with multiple meanings, and more meanings when context is different, and a machine can't possibly understand all of that. The entire concept of "engrish" is a prime example of how even individual words and phrases can be mangled, let alone long-form content.

AI has been throwing a bit of a curveball here. It's better than a lot of the previous kinds of machine translations available. But is it good enough yet? I'm skeptical.

On one hand, I've been seeing more and more articles that claim AI has reached human-level translation quality. I've even been getting a lot of emails recently from ChatGPT encouraging using their system for translation.

On the other hand, all of those have come from companies selling AI translation, so I'm hesitant to take them at their word.

It's also extremely dependent on the language. A recent study found that AI was "noninferior to" (so, on par with) human translation for Spanish, but worse for other languages like Chinese, Vietnamese, and Somali.

I think for some basic kinds of content, especially formulaic or very simple kinds of content, AI translation can work fine. But if you're translating an entire website, you have a lot of different kinds of content to publish, and you definitely can't blindly trust AI with all of it.

The biggest issue, though, is that unless you're bilingual yourself, you have no way of knowing. You know all those pictures of people who get cool tattoos in Chinese, never knowing they got "dishwasher" permanently emblazoned on their calf? Yeah, it's like that.

You have three options here.

1. Trust the AI and hope that if there are issues, someone points them out, and they don't cause you problems. This is the worst option.
2. Hire someone who is either bilingual or a translator themselves to review your content and validate its accuracy.
3. Hire a human translator to do the job right the first time.

Option 3 is the most expensive, but option 2 takes longer, since you have the initial back-and-forth, and then you have to do something to get more accurate translations when it's reviewed. 

I will say, trying to hire someone to "fix" an AI translation misunderstands how complex translation is. I've talked to a lot of translation specialists over the last few years, and one of their biggest complaints is that "fixing an AI translation" usually just means starting over and translating from scratch, since there's no fixing the kinds of mistakes AI can make. If you're looking for a good place to hire a writer, it's worth considering specialized platforms rather than generic ones.

All of this can vary depending on the kind of content you're translating, so it's up to you to make the final choice. Generic sales copy? Maybe AI is fine. Tech specs, legal documents, or emotionally resonant storytelling? Get a human.

Forgetting to translate SEO information.

Another big mistake a lot of businesses make is thinking the basic content on the page is all they need to translate. There are tons of website plugins to automatically translate some elements of a site, like your navigation buttons, but that's just one part of the puzzle.

There's a lot that goes into SEO that you might lose if you're not translating it. Things like:

• Your meta title, if it's different from your H1.
• Your meta description, if you aren't just letting Google generate their own.
• Your Schema markup; the properties are English, but the content you fill in can be localized. It's also worth knowing whether you can use multiple schema tags on the same page.
• Your image alt text.

Even your URL might benefit from being translated. That, however, depends on how you're displaying translated content, which is another point I want to make.

Not Making a Separate URL for Translated Pages

I've seen some sites that have a language selection drop-down and use a script to dynamically translate the content of a page, either using a plugin or by populating with prewritten translated content, though the automatic translation is a lot more common.

This is a terrible option for SEO.

SEO relies on URLs and indexing the content of those URLs. When you use a system like the above, you're effectively hiding your multilingual content from search engines. As far as Google is concerned, you don't have that content on your site.

Some sites use a subdomain (www.fr.example.com), while others use a subfolder (www.example.com/fr/) to organize their alternate-language content. Either one of these can work, but usually a subfolder is preferred. You don't necessarily need a country-code TLD (www.example.fr) but, for larger sites, you might consider it.

Make sure to use the hreflang meta attribute in your headers, so search engines don't have to guess at the language, since they can get it wrong occasionally.

As a side note, Google also recommends not automatically redirecting a URL based on user attributes. Just because someone is browsing from France doesn't mean you need to serve them a French version if they landed on the English page. They might be a vacationer or expat!

Failing to localize relevant data.

This is a minor detail, and more focused on commerce sites, but it's still important: translate your prices!

If you're serving your page to people in Europe, show prices converted to Euros; don't keep it in dollars. This is something that can be done with a plugin, so there's no excuse.

Ignoring language in images and other media content.

I generally prefer to avoid putting text in images in the first place, as a holdover from when Google was bad at indexing image content. If you put text on images, or if you embed videos or other forms of media, translate that too. Make sure you also handle your image alt text and captions for each language version. 

Yes, it means you have to make more media, and it adds a storage burden and maintenance burden, but you can't really ignore it. These are key to your site, so leaving it out can show you don't actually care about translation.

Not adapting your website to alternative language displays.

A lot of languages are like English in that they're read from left to right, but not all of them. Some languages are read right to left, so if you're going to translate your page into those languages, you need to adapt the way your site displays to suit those languages.

It's a small detail, but it really matters to those audiences, so it's an important one to get right.

Assuming translation is all you need to open a new market.

A lot goes into doing business in another country, and I'm not just talking about logistics like shipping. Think about the legal and regulatory environments. Think about regional customs. Think about not saying "I'm a jelly donut" in front of a crowd, you know?

This is a matter of localization versus translation. Translation is a version of the text that keeps the same meaning; localization is a version of the text that can change the meaning to suit the same goals and emotional resonance of the target audience. Localization is by far the harder thing to pull off, but it really can make a huge difference.

The Right Way to Translate Your Content

Avoiding problems and pitfalls is important, but it doesn't get you across the finish line. What do you need to do to properly translate your content, without hurting your SEO?

Follow Google Best Practices

I already linked Google's page, but I'm doing it again.

Here are the key points to remember:

• Use different URLs for different languages/regions of site content.
• Use meta tags like hreflang to identify the language on a page.
• Make sure there's an indicator on the page of what language/region the page is for. Do more than just a little flag icon, too.
• Have an easy-to-use way for a user to change the language.
• Don't automatically redirect a user based on what you think they want; let them choose.
• Use language-specific URLs, but properly escape them into UTF-8 instead of using alternative character sets.

There are also two key details that you might want to know. One is that Google always crawls from USA-based servers. They don't try to fish for alternative versions from foreign IP addresses.

The other is that they ignore certain meta tags that you might use for identification, like geo-position or distribution elements. These have been deprecated for a long time, but some folks still discover them and attempt to use them, to no avail.

Finally, don't forget to canonicalize the primary version to avoid duplicate content issues. You usually shouldn't run into this with translated content, but it's better to be safe.

Invest in Localization When Possible

Localization is a lot harder than pure translation, but it's also a lot more effective, because it really shows that you care about trying to present information in the language or to the culture you're creating your regional variation for. 

Even things like your base-level keyword research can be very different. The kinds of terms people use in other languages to find your content can vary. Your product category, your niche, even your industry can be referred to with entirely different terms.

Localized keyword research is hard! It's understandable that many businesses don't do it. But, if you're serious about operating in a secondary language, you really need to treat it like a second primary, not a second-fiddle imitation of your main site.

Periodically Review Translation Accuracy

I have a whole huge section above about how AI might not be entirely trustworthy, but you'll notice that I didn't say to avoid AI entirely. There are two main reasons for that.

The first is that AI can do some of the job just fine. It can make mistakes, get things wrong, and mistranslate things, but it's getting better, especially for certain languages.

The second, though, is that humans are also fallible. If you're paying a human to translate for you, they may or may not be doing a good job. These days, they may even be using AI and hoping you won't notice. 

Trust, but verify. It can be worth periodically hiring someone to review your translations for accuracy, point out major problems, and allow you to hold anyone you've paid accountable for their translations.

Pick Your Target Languages Carefully

I think one of the biggest mistakes a lot of companies make when deciding to translate content is picking as many languages as they can and hammering out translations, often using MTL or automatic plugins, and calling it a day.

I don't think this is worthwhile. 

The main reason is simply that most of those languages aren't going to be bringing you traffic, and your bare-minimum translation isn't going to change that.

I generally recommend starting with one, or maybe two languages, based on the existing interest in your site. If you know a lot of your customers speak Spanish, work on a Spanish version of your site. If you know you have a sizable population of Somali immigrants in your community, consider hiring a local to help you create a version of your site for them.

It's also important to understand why you're translating what you're translating. Like, for example, I wouldn't benefit from translating my site into other languages. I provide content marketing in English, and don't currently have multilingual writers on staff. What good would it do for me to offer a Spanish or French or Arabic or whatever version of my site? Anyone from those regions interested in my services would be interested in my English services anyway.

Put The Effort Where It Matters

Another common mistake is mindlessly translating everything. 

That two-year-old blog post that hasn't seen a visitor in six months? Yeah, you don't need to translate that. At least, not right away.

Keep the 80/20 rule in mind. 80% of your results come from 20% of your effort. Find the top 20% of pages on your site and translate those. IF that proves to be valuable, you can expand your translations and localize content as you go. Otherwise, maybe don't worry about it.

Another consideration is your publication schedule. 

You have to write content before it can be translated. But if you're on a tight turnaround with your publication schedule, you might not have time for translation. Is it okay to publish your translated content on a delay?

This depends a lot on what content you're translating. For standard blog posts, sure, the delay won't hurt you. Some users might feel miffed that they're on a delay, but it's less common than you might think. 

On the other hand, if you're trying to follow current events, announce sales or promotions, or do anything time-sensitive, you want more of a lead time so you can establish your translated content to publish simultaneously. 

Overall, translating your site and getting it to function effectively in another region is a big expense in both time and money. There's a reason most businesses have either ignored it or gone with low-effort, mistake-ridden translations and just dealt with it. AI is helping those winds change, but we're still a ways away from effective AI localization, so all of the usual considerations continue to apply.

Do you think it's valuable to translate your site? Have you gone all-in on regional sites? How has it worked out for you? Let me know your stories.