Over 100 volunteers have translated over 300 tutorials and articles into over 25 languages for the Tuts+ Translation Project. That’s remarkable, and I feel that this response from our community alone proves that the project is a worthwhile initiative. However, as a data analyst, it’s my job to put my feelings to one side and explore how it’s doing using numbers and graphs.
Click on a translation, and the countries change colour. The darker green a country is, the more visitors the post got from there compared to other countries. The map also shows this information for all translated posts on aggregate.
We can see that different languages of post attract different geographical profiles of readers. This might not be surprising, but it’s a very important result: it shows that the translations are reaching different areas of the world, not just people living in English-speaking countries that would prefer to read posts in a different language.
I’m also interested in how people are getting to the translated posts. We can categorise visits into two main paths: those that come from inside Tuts+, and those that come from outside Tuts+.
In this graph, the yellow line represents all pageviews across all translated posts that the reader got to by clicking through from the English version of the post. (This is the only way we would realistically expect readers to navigate to translated posts from within Tuts+, as we don’t yet have language-specific index pages.) The grey line represents all pageviews across all translated posts that the reader got to from an external source: a link on another site, a tweet, a Google search, a bookmark, or similar.
It’s clear that far more people get to the translations from outside Tuts+ than from inside. This might indicate an issue with visibility (that is, many visitors might not be aware that so many posts have been translated into their languages), or it might simply be the case that people that prefer reading in languages other than English don’t tend to visit Tuts+ (yet).
Let’s break this down further, and see which external sources bring people to the translated posts.
June’s spike in social media traffic is due to Ian publishing translations for a post that he already knew was very popular, and then promoting it heavily on Facebook and Twitter; it obviously spread! August’s drop in social media traffic is due to Ian deciding to stop promoting translated posts on Twitter and Facebook after followers complained that these posts weren’t relevant; this is a fair criticism, of course, but it’s a pity that we don’t have a better solution, as obviously a lot of people did find the translations relevant.
It’s very reassuring to see that search traffic is steadily increasing. If people only found translations through the site or through our active promotion, and never organically through search, that would be a bad sign.
Here, the red line again represents total search traffic to translations, and the grey bars represent the number of translations on the site. We can see that there’s an almost linear relationship between the number of translations we publish and the search traffic we get. It would be great to see more of an upward curve in that red line, but even just seeing the existing trend continue would be a good result, as there are plenty of tutorials left to translate.
It’s also useful to break down the traffic landing on the translations by the language of the translations:
Spanish translations are in front by quite a margin, but I believe a lot of this is simply because we’ve published far more translations in Spanish than in almost any other language:
We can see that there are about half as many French translations as Spanish ones, and they get about half as much of the traffic; there are roughly half as many German and Russian translations as French ones, and again they get about half as much of the traffic. But this pattern doesn’t hold for all of the languages: Portuguese and Indonesian are obvious exceptions.
We can chalk this up, at least partly, to the fact that there are fewer Portuguese speakers than Spanish speakers, and fewer Indonesian speakers than French or Russian speakers. The selection of posts that have been translated may also be a factor. But in at least one country’s case, there’s more to it than that…
If you’ve explored the interactive map, you may have noticed that clicking “Bulgarian” makes Bulgaria light up bright, “Hungarian” makes Hungary light up bright, “Swedish” makes Sweden light up bright, and so on, usually with a few other countries lighting up dimmer as well. Click on “Chinese (Traditional)” or “Chinese (Simplified)”, however, and China does light up, but only dimly—no more than any other country on the map.
As Kendra Schaefer explains in Notes From Behind the Firewall: The State of Web Design in China, it’s not easy to get your website seen in China if it’s hosted outside the mainland. This graph puts that in perspective:
That’s another graph showing the number of people coming directly to the translations from external sources, using the same scale as the “Pageviews for all translated posts, split up into specific external sources” graph, but this time it’s filtered to show only the Chinese translations.
The total traffic is low (hence the whitespace), yes, but the main point is that almost all of the traffic to our Chinese translations came from social media—and almost none from search—so when we stopped promoting translations on Twitter and Facebook, it became very unlikely that anyone would find the Chinese translations on their own.
Our tagline is “Teaching skills to millions worldwide”, and the Tuts+ Translation Project is helping us do that. The results we’ve seen so far might not be surprising, but they are promising and reassuring.
Having said that, this analysis has revealed several areas for improvement: making our existing visitors aware that we offer tutorials and articles in different languages; keeping regular readers updated on new tutorials and articles written in their preferred languages; and making headway in countries like China where we don’t yet have a foothold.