Perspectives: Going global via the cloud

We’re surrounded by cloud-based services today, especially in our non-work lives. We use Gmail for personal email. We stream music from Pandora and Spotify. We purchase books, and just about anything else, from Amazon on our phones or other mobile devices. We watch movies and TV shows on tablets. And it’s all done through the cloud.

Small businesses are basing their whole technical infrastructure on cloud-based services too, from accounting and relationship management to marketing and advertising. Enterprises are relying more and more on cloud-based services as well — they are creating “internal clouds” within company firewalls and datacenters for internal access only.

The term cloud-based really means that you can use software and storage as a utility: you pay for what you use, rather than building the infrastructure yourself. Instead of laying in the wires for your home telephone service, you buy a smartphone. Instead of connecting to cable TV, you subscribe to a video streaming service. So what is the difference with translation management?

Let’s take a trip back in time. Come with me; I was there. In the 1990s, we, meaning the people who were thinking about translation back then, were starting to use translation memory (TM). It was a relatively new concept, and it wasn’t too popular with translators. I remember in my first translation management role in the late ‘90s talking to the translation agency my company had hired, and asking why they were not using TM for the translation of our first company website. It was a website for customers who had a support contract, and it was being translated into Japanese. There was very specific terminology to keep track of, along with regular, small updates to each page. The development team was providing the source files plus a UNIX file that indicated the number of changes since the previous version — just the total changes, not where to find them! And yet, the language service provider (LSP) was refusing to use TM. I was horrified. Why? What was the excuse?

The LSP told me, “there’s too much overhead involved. We have to load the files, run them through the TM and then send them to the translator. We might as well just send the files to the translator and retranslate the whole thing.”

Remember. This was before 2000. However, it seemed blatantly obvious even then that there were tasks in the end-to-end translation process that could, and should, be automated and made mandatory. This included the use of TM, regardless of the size of the project. It’s not only a matter of cost or efficiency, it’s a matter of the company voice, brand and style. It’s important!

At the time I was part of a huge enterprise company. While that may suggest that we had unlimited budget and unlimited resources, believe me, we had zero! With one colleague, I researched and investigated what little technology there was available at the time. Yes, there were desktop TM tools for translators. But there was zero automation for the rest of the tasks. This meant many things were far too manual, such as extraction of content, conversion of file formats (UNIX used by developers, MS-DOS by translators, and yes, we needed to worry about that), identification of component files in a website by reviewers and testers (we had to treat a website just like a software product) and so much more.

We discovered a small startup called Uniscape that had created translation workflow with TM and terminology. We negotiated a pilot and launched a do-or-die project: one language, one release with this newfangled translation workflow technology. If we delayed the release of the Japanese website, it would have been… well, it would have been one of those moments you remember throughout your career.

Happily, it worked. There was definitely some behind-the-scenes “sneaker-net” going on, but for all intents and purposes, we had created and piloted an end-to-end, automated translation workflow with integrated and shared TM, terminology, human expert translators, in-country reviewers and a (still-too-complex) process for linking the quality assurance process back to the component files.

Over the years, the whole infrastructure grew and morphed from an externally hosted and subscription-based service at Uniscape to a huge, company-mandatory and internally-hosted, datacenter-based system — though still managed by an incredibly small team of experts. Uniscape? They were acquired and acquired again, now part of SDL.

It was mandatory because we were easily able to show our C-level executives that there is a very strong return on investment for centralized translation management. Not only is it the most efficient way to manage translations, it also enables companies to quickly reach new markets. It’s not a cost, it’s a revenue enabler. It’s something that each and every company should be doing.

It was a similar story in the other huge high-tech companies. We had a small clique of peers, and we’d get together and compare notes at industry conferences. We talked about what we’d done, how important it was for global business — and we thought everyone else was doing the same.

Ooops. Not true. So who was doing it? A few teams or individuals with the vision to see that centrally-managed computer-aided translation technology is an essential part of the enterprise, and a few high-level management teams with the insight and trust in their experts to invest in the technology. Everyone else is still, even today, mostly using desktop TM tools, very manual content extraction and delivery processes, black-box LSP translation or technology services. Even some huge manufacturing companies still have siloed and, in my mind, archaic translation processes. They are stuck in the 1990s, heads down, focused on not missing the deadline for that translated website — or product manual or marketing campaign.

And yet, look at what the new companies and emerging brands are doing! Companies that have blossomed because of the cloud, because the cloud makes it possible. Talk to any CEO or employee of a cloud-based business, suggest that they use desktop translation tools, or manual extraction of content, or translate “blind” with no context, and they’ll laugh you out of the room. Like offering a landline connected through an operator at a local telephone exchange to someone who’s already seen a smartphone: it’s laughable.

Today, there is no excuse. You don’t have to be a huge enterprise to have efficient, centralized translation management technology with the ability to share TM, terminology and style; to automatically detect and extract content changes; and to disperse content to every actor in the translation process. You don’t have to be a huge enterprise to simplify your translation work, nor to easily grow into new markets. You simply don’t have to buy or build or hire the infrastructure, the datacenter, the expert team. Every company, no matter the size, can go global. We indeed live in a cloud-based world.