Unbabel (YC W14) raises $60M Series C to become the world’s translation layer
techcrunch.com> We are now translating over 1,000,000 customer service messages a month
> The company told us it now has more than 100k registered translators providing translation services for its platform.
So... each human translator translates (on average) 10 messages per month. In fact, less than 10, because (presumably) some messages are also translated by AI.
Odd numbers.
“100k registered translators” probably means “5-10k active translators” (just a guess).
Either way, translating 1mm customer support messages per month sounds impressive regardless of the number of people translating.
"TL;DR: There are about 100K-200K unique workers on Amazon. On average, there are 2K-5K workers active on Amazon at any given time, which is equivalent to having 10K-25K full-time employees. On average, 50% of the worker population changes within 12-18 months. Workers exhibit widely different patterns of activity, with most workers being active only occasionally, and few workers being very active. Combining our results with the results from Hara et al, we see that MTurk has a yearly transaction volume of a few hundreds of millions of dollars."
https://www.behind-the-enemy-lines.com/2018/01/how-many-mech...
I've known Vasco since we met in the Pittsburgh tech scene a decade ago. Such a great guy and been through a lot to get here. Congrats Unbabel!
Congratulations to Unbabel, but I've heard some nasty horror stories about the inside and I would be skeptical of any claim to become the "world's translation layer".
However, if investors are willing to risk it, then good luck to them. More Portugese tech companies getting funding is always a good thing!
> * I've heard some nasty horror stories about the inside>
Is this related to their technology, their culture, or both?
Not OP, but I've heard they used to have a compulsory surf day per week or month, or something like that. Not bad if that's your thing..
Among other things, europe's language diversity is listed as one of the reasons why europe isn't a good place for founding a tech company: the american market is much more homogenous. However, for translation startups, this diversity makes europe the best place to be in the world.
They definitely won't run out of diversity, and it also has the european single market which makes it easy to conduct business across inner-european borders. Furthermore, the single market is only a recent addition so many companies and governments which start having to communicate across language barriers are still figuring out how to do it. Canada for example already has an established system as Switzerland has, too. But on an european level, the market is new and fresh and there is no incumbent that needs to be disrupted. It's great to see unbabel and deepl choosing europe as their main base of operations.
Another way of looking at Europe's language diversity is that having to deal with it early on will make you better prepared to go global.
The American market is homogenous but crowded. World markets are increasingly open, but Americans aren't necessarily best positioned to take the lead.
> having to deal with it early on will make you better prepared to go global
Do we have evidence of this?
(European start-ups do tend to have a competitive advantage with administration, given the complexity and diversity of regulatory regimes on the continent. This explains e.g. Spotify (negotiating music rights) and Klarna (negotiating with banks).)
Europe's language diversity is a trivial cost of business and does not impede tech companies.
But, obviously, that does make Europe a very good market for translation companies.
> Europe's language diversity is a trivial cost of business and does not impede tech companies.
This can't be serious. This is laughable by any stretch of the imagination.
Here's a list of how it impedes tech companies:
1. Localization effort for 20+ languages (for mainland EU). Every time a new feature that has some words included in it, it has to be translated. Screenshots in support have to be updated.
2. Training - if you have to train users, you need language specific training.
3. Account management - account/success managers need to be language specific.
4. Support - having support personnel that can speak different languages.
None of those are trivial and all cost significant money.
It is indeed laughable to claim that absolutely all languages should be supported at all times and that a localisation is costly ("screenshot have to be updated", come on...) By the way, in the EU a population equal to the US's is reached by going for the 5 most populous countries (granted, as it happens that does mean 5 languages, including English, but that also means that these 5 languages cover a population larger the US's).
The rest of the points, if needed, grow with the size of the customer base and do not impede tech companies' growth in the EU v the US.
Rudeness and bluster are not arguments.
Clearly your definition of "impede" is very different from mine.
> should be supported at all times
Where did I say that? We're talking about impeding growth, which is along a continuum of time. Any additional "n" amount of effort required during a growth period is going to slow it progress. Additional effort is absolutely necessary to support a different local language by definition. It doesn't mean growth is impossible, it just means it requires extra effort and resources, which in turn means additional time and money.
> that a localisation is costly
Are you claiming it doesn't incur any cost? At what scale (let's use lines of translated text as a barometer) have you ever operated at?
> ("screenshot have to be updated", come on...)
Have you ever had to create support documentation in different languages? For the French specifically? Your reaction tells me this is not the case.
> bluster are not arguments.
Where specifically was I acting in a threatening way in my argument?
Canadian here and doing this for French and English has not been a major deal breaker. One of the qualifications we put on our documentation team was being bilingual.
You can also NOT use typical screenshots for documentation but rather have templates screenshots that contain no text and then layer the text over top of those and use things like language selection in the browser to load the correct language.
Employee training can be simplified by headquartering out of one country and having a "primary" language that you work out of day to day.
For many early startups having full fledged customer support (call center style) isn't likely anyway, doing it via chat / email / support tickets can be translated or you can diversify your support hires to cover the languages you officially support.
So is there additional costs - definitely - are they enough to impede and side rail a startup... likely not, and if they are that startup likely wouldn't have been able to succeed anywhere else either.
You're getting push back for the same reason you are shaking your head at the other poster - they are making it sound like it's free, and you're making it sound like it's going to crush someone.... it's slightly more effort with possibly a greater reward if you are able to do it properly.
I acknowledged that this was a cost and never suggested it was free.
My point is that, in general, it is a small cost relative to everything else.
Another point that someone in the US might not know is that if you are in a major EU capital city or around you can relatively easily hire people who are fluent in at least 2 languages, so some level of multi-lingual support may thus come "for free" and it's not unusual to have 4+ languages spoken at native level even in a relatively small team (in the tech industry).
True, and you left out a whole host of other issues.
While EU regulations harmonize a lot, it's still much, much messier than in the US.
* different tax regimes
* often quite different regulations
* contract requirements
* income levels which can affect pricing decisions / require regional pricing
* localized ad campaigns per country
* getting media attention in each country is hard
* ...
All of the exist in the States? I'm not saying th EU is easy, but state legislatures have made it increasingly hard.
> Europe's language diversity is a trivial cost of business and does not impede tech companies.
Are you suggesting that the language diversity doesn’t impede tech companies from reaching English speaking markets the same as American tech companies do? Or that the language diversity doesn’t impede companies from reaching those language-diverse markets?
Do you have some reasoning or evidence to back up that dismissal of a very plausible concern?
Or do you think it's coincidence, or due to economic structure, that the US and China dominate the tech scene?
This is a pretty widely understood hurdle for EU startups, and it's unclear why you dismiss it out of hand.
It is often repeated, but does that make it true or understood?
My reasoning is that if you're a tech company then translation costs are trivial (that's close to obvious)
Now, depending on your industry, cultural differences may be an important factor but that's different. All the well-known tech companies are equally present across the EU so it does not seem like a major problem (that may be different across widely different cultures).
Even large companies like Amazon or Facebook launch some of their new products only for the american audience and only expand to europe later. And early stage startups don't have the resources of translating their app into two dozen languages, offering customer service in those languages, having a sales infrastructure, etc. It's doable but more expensive if you operated in the american market.
What does that have to do with language?
Now, if you're an early stage startup you have your hands full already, you can wait until next year to spread across the EU in different languages. That does not impede anything
For support, it's quite easy to hire people who speak different languages and the total cost will be the same.
> you can wait until next year to spread across the EU in different languages. That does not impede anything
It literally is impeding you from operating in that other country right now. What do you not understand about that?
Take a like for like example:
I have 10 customers in Texas who use my software and they pay me $100k/year. I can sell my software to 5 more customers in California...tomorrow. So I make $150k this year.
I have 10 customers in the UK who use my software and they pay me £100k/year. My software isn't localized and therefore I can't sell my software to 5 French companies (because they require the French language for their employees). I have to wait a year to earn the additional £50k/year while I add support for them. I only make £100k this year.
I just lost out on an additional £50k. How is that not an impediment to my growth rate....it's literally 50% less...
Oh, I also have to hire French people to support them which increases my costs and therefore making me less profitable.
impede =/= prohibit
Do you have actual experience building a product with cross-language support, or are you hypothesizing?
I do have experience.
For companies the size of Google, it's trivial. For startup companies, it isn't. Translating the interface is comparatively easy but if a large component of your bug reports is in a language you can't understand you have a problem, the same problem that translation startups are trying to fix :).
Not sure why you would bet against Google on this.
Well, even Google is partnering up with Unbabel, so there's potential: https://theworldbreakingnews.com/unbabel-announces-partnersh...
Google translate is tolerable for reading content in a language you don't know, but it would be terrible for translating messages to your customers! Human-human interaction requires a lot more understanding of nuance and idiom and even things like culture-specific expectations of politeness. Translation AIs are nowhere near good enough.
Even when you are interacting with a human support agent they're usually reading/copy/pasting from a script, noticeably jaded and lacking empathy from dealing with a never-ending stream of complaints that they are not empowered to actually do much about, and/or not a native speaker. There's not usually a whole lot of nuance in these interactions as it is.
Reminds me of back when Google Translate used to be called babel. Did Google acquire them or was it just a branding change?
there are already excellent suctions to this problem... why is this better?
and why do you need/want ai to be involved?
Please run as fast as you can. This company is toxic waste.
Can you explain why you hold that opinion? I’m curious as two people have mentioned it in the comments so far.
That same user posted this “Tell HN”: