Website localization should be straightforward. No worries here, content is still king, independently of whether it is written in English, Spanish or Chinese. So if you provide content in different languages why should your fear the mighty Google and its algorithm changes? Your multilingual SEO efforts will pay off, and you will survive in the SERPs. Or will you?
In January, Google promised that it would take action against content farms that were gaining top listings with “shallow” or “low-quality” content. Said and done. Google rolled out in February the so-called “Panda” algorithm (also known as “Farmer”), which impacted 12% of its search results in the US. This is a far higher impact on results than most of its algorithm changes. Though the change only impacts the results in the US, it is likely that be rolled out worldwide in the near future.
Laura Lippay explains in her interesting Google told you so post that Google is getting rid of low-quality content, and that this has being increasingly the case. Ajayseo in Digitalpoint provides some interesting tips to survive the algorithm change, which come basically down to improve the quality of the content. And every time you start looking for information on this algorithm change, you get the same message: Google is coming hard onto the low-quality content.
So what has this to do with website localization, you might ask? Well, obviously that Google will sooner or later (probably the first) roll out this algorithm to the rest of the world. And your pages in Spanish, German or Chinese, which rank so nicely at the top of the local Googles, might suddenly shift into oblivion. Why, if they have useful content? Well, because of the way that Google detects spammy sites.
Now, nobody nows exactly how Google identifies the spammy sites, but we can make an educated guess. For example, spammy sites usually take to article spinning, so as to show a lot of “unique” content. Whoever has performed article spinning, however, realizes that very often the constructs of the spinning is not exactly how people would write. Spinners replace words by synonyms, or parts of the sentence by a different one. Speech is often forced, grammar is questionable. Another element that characterizes spammy site is a great number of low-cost articles, usually written by low-wage workers, often in third-world countries. Grammar is often dubious, if not outright wrong, there are spelling mistakes… that kind of thing. Would you believe a high quality site is riddled with these things? I doubt it. And Google, which has probably the greatest database in the world of information, can very easily distinguish misspellings and grammar errors. You don’t believe it? Then check the article from Matt Cuts about that Google quality is hiring – the second sentences says it all: “We work on everything from synonyms and spell correction to core ranking, UI changes, evaluation, and yes–even webspam”.
There will be other signs, of course. But grammar and spelling are likely to be some of the parameters for the flagging of these sites, or Google would have rolled out the new algorithm immediately all over the world. The fact that it didn’t do it is a big flashing light that word recognition has something to do with it. Still skeptic? Then check out my post on how search engines identify language. Search engines are much smarter than you think. Several discussions in webmasterworld also highlight the issue of grammar and spelling , with one user reporting that he dropped 400 positions because of one single spelling error in an <h3> header tag. The good news is that he bounced back within days after he corrected the error.
And here comes now why you should potentially be affected by the algorithm changes: A lot of the website translation and localization is performed with low-cost translators and/or machine translation. So what is the result? Translated pages are in these cases likely to have grammar errors and typos – a big red flag. Check out for example my post on machine translation quality. Now, as these things are quite easy to detect for a mammoth like the big G, which uses already dictionary and probabilistic methods for language recognition, your translated site -no matter how much original content you may have- may be inadvertently flagged up as search engine spam, instead of a genuine website localization.
Now, if the mighty Google indeed uses things like grammar correctness and spelling errors to get an idea about the quality of your website (as Matt Cutts seems to suggest), and you have resorted to some cheapy website localization for you multilingual SEO, then you are in deep trouble. Such low-cost translation might eventually cost you a very high ranking of your localized pages. Your international SEO efforts will be wasted due to such a stupid thing as saving on the wrong thing. Again, it is evidenced that penny-wise, pound-fool. And that cheap in the long run becomes expensive. But you might be still on time to rescue your web localization efforts, it will take still some time before Google extends Panda to the rest of the world.
Mind you: You were warned!