This is it: another great experiment that will change the way we build content websites forever.
Obviously from the title, we’re translating the hell out of our million-word case study website.
So no more introduction. Let’s get into it.
Look at these sites
Now, the bottom site is my site. The top is a site in the exact same niche…but in Spanish. That’s what triggered the idea.
After doing a bit more digging, I found a few fantastic keywords with damn near no competition.
Thus…
The Great Translation
All it is is taking my existing content and translating it to
- German
- Spanish
- French
- Portuguese
- Italian
- Polish
- Dutch
- Russian
The problem is
- I don’t speak any of those languages
- Translating articles is pretty damn expensive. I’ve tried it in the past, and spent about $875 for 12 articles/15k words to be translated
- And because I don’t speak those languages, I can’t gauge if the translation is good or not
So it’s tool time 🔨
Yeah, we’re gonna go the fierce route because we are on a budget, and tools work pretty damn well (enough). Enter: DeepL & Neuronto.
DeepL is the translation tool with API access, Neuronto is a plugin that connects to the API to automatically translate articles.
Eventually, perhaps, maybe, if I feel like it, I’ll hire a bilingual VA to go in and make sure the articles read appropriately. In the meantime, however, it’s publishing time.
The problems thus far
Silly me tried to auto-translate ALL posts to Spanish (twice), and bricked the site for 42 minutes. I learned the hard way that we’ll need to go in and manually create and edit the translations.
The plugin doesn’t read reusable blocks, so for the time being, all the verses and alternatives-to articles will not make the cut. Over time I’ll need to convert reusable blocks to just blocks, but that’s a massive undertaking.
Neuronto is pretty damn wonky with Gutenberg blocks, so I have to
- convert reusable blocks to regular blocks
- convert posts to the classic editor
- clean up extra code on the html-side
- translate and clean up even more converted crap
- convert original post back to blocks
Additionally, non-generic type blocks are fucked, because we have to convert to classic posts. So all the sexy buttons and containers and columns get messed up properly.
On top of everything else, the million-word case study is more like 5 million words now. We’re essentially multiplying our largest articles by eight.
But it’s not all bad
At this point, I can go through older posts and clean them up, add more information, update content, add rich media, and add internal links. It’s more like a mass content update and audit.