Hello? Translation industry? It’s 2020

by | Dec 9, 2020

Time for an agile approach

In this article, I will argue that our current ecosystem reinforces translation to be treated as separate from content creation, leading to a lack of innovation in the translation industry. This is the reason we have yet to solve the problem of publishing content in multiple languages. It is still slowing down the time to market, leads to numerous quality issues, and is too expensive to guarantee a proper ROI.

How it all began

When content was mostly published on paper, you almost had no choice but to take a waterfall (avant la lettre) approach to translation. Translation management was a tedious task, mostly because content was only stored in files, and was sometimes even shared through faxes and other means of hard copy devices.

Making a change to something that was already sent out to a translator was a no go. So a source text had to be fully revised before it could be sent out, and if there was a necessary change, this was a big deal for whoever was in charge of getting translations done.

Of course, no company wanted to deal with this. Because who wants to add to their core business:

  • Finding translators by putting adds in newspapers and getting through tens or 100s of applications
  • Sending documents back and forth to translators
  • Making changes in the source text in 10 languages that were corresponded through email
  • Doing all the tedious work that project managers in LSPs are doing these days (and that makes them cry on a regular basis)

This is how translation companies came into existence. Yuka Nakasone once told me the story of how the entire translation industry — including all LSPs, MLVs and SLVs — was designed and ordered by a tech giant. With their hundreds of millions of dollars of localization projects back then as well, they were able to order the entire ecosystem as we know it right now.

Translation companies promised to fix the problems they encountered in the translation process, even if they were caused on the client side. They optimized their part of the system, starting off from a signed off source text, and maximized their revenues on that model. Very useful! Until now…

Moving to an agile mindset in an online world

In this digital world, time to market is everything. The sooner you publish or deploy, the sooner you can start making money. Holding off on publishing until everything is perfect is a great pitfall and can destroy the value you create.

All content is online and can — and will — be changed on the basis of user feedback at any time. People writing content — whether they’re in marketing, support or in product, know that they’re not almighty and that they need to adjust their content on the basis of the wisdom of the crowd.

Effectivity of content is key, and whether or not it is effective can be measured in one way only: by giving it to the user. But it needs to be fast. And to reach the full potential: it needs to be fast in all languages.

Agile methodology in software development and other industries, has also taught us that having multidisciplinary teams collaborate in creating MVPs (minimal viable product) and iterating over that first version, is far more effective than working in silos, where there is no way “up the waterfall” and problems detected later on in the process can never be fixed where they originate.

When we make the mindshift to a more agile approach to translation, we start thinking differently about a lot of things. Semir Mehadzic, one of my fellow agile minds in the localization industry gives some examples in reply to one of my posts on LinkedIn:

  • We don’t want to do LQA (Language Quality Assurance, aka reviewing — ed.) in the same environment/context as the translators
  • We’re not even sure separate LQA is needed at all
  • We’re OK with the possibility of releasing an occasional spelling or other error to production (to save time and reduce the complexity)
  • LSPs (aka translation agencies – ed) are really not needed if you have text strings to translate and have an automated way to dispatch them
  • We do not believe in KPIs and error-count metrics as a way to measure quality

While these statements were absolutely unimaginable 20 years ago, the current situation of how online content is stored and the requirements of the world we are living in, are asking for these kinds of bold statements. That is, if we ever want to deliver a high quality multilingual product – quality as defined by the user: what he needs, at the time he needs it, at the costs that he is willing to pay for it.

And, of course, this online world also makes it easy to find really good translators, that know how to translate, but also, that understand much more of the translation process than you can ever imagine. I was shocked when only 4 years ago, at a conference, I heard people saying about translators: “oh no, they just want to type words on their attic, they don’t want to be bothered with client side problems, nor be involved in the process.” Some waste of talent and expertise, right?

We can store content online before it is published, we can make and track changes at any time, we can find translators online, we can get feedback from our users. We have project management tools in the cloud, we are able to track what has happened when. We can see lean and agile processes implemented in other industries, all we have to do is look at what they do and what they’ve learned.

The time is now! We have all the ingredients for real collaboration and the right use of talent and expertise, and we can finally eliminate some serious waste in our global content processes. We have all the tools to really become creative and design more sophisticated processes. Finally, we can give equal rights to speakers of all languages and serve them with content in their languages without a blink of the eye!


Eh… no

Where we really are

In most companies, translation is still considered a cost instead of an opportunity. Many companies still do not translate their content because they fear the costs and the time it takes, which is of course negatively affecting their international ambitions. In case translation is felt to be necessary, it usually leads to a long delay in time to market, for both content and products.

You could even argue that world economics are not reaching its full potential because we cannot overcome the language barriers and we are not taking full advantage of our product potential internationally.

Somehow, we are not using all the opportunities this new digital world is offering. Translation is still an afterthought for most companies and this is making the translation process run into all kinds of bottlenecks.

In a way, it is not surprising this is happening.

The ecosystem that currently exists is inherently set up as a waterfall process. In many cases, there are three processes that function independently from each other: the content creation process, the in-house project management of translation and the outsourced project management of translation.

You finish your source text, you notify your translation department, who sends it to a translation agency, they do something with it (potentially through subvendors, all with their own process) and then a translation is returned. The translation department is responsible for making sure the content is validated and published.

There is usually no contact between the translator and the owner of the content, there is hardly any feedback about textual or technical issues with the source text, and if there is, there is no system that allows for these kinds of feedback loops, as is usually the case in waterfall processes. All this while from the translation process, valuable information can be acquired about the quality of the entire process of the creation of content.

Why is this a bad thing?

One of the main principles of Lean is to optimize the whole: optimizing the entire value stream. Suboptimization is “the practice of focusing on one component of a total and making changes intended to improve that one component and ignoring the effects on the other components”. Suboptimization is a problem, not only because you often optimize the wrong component — which does not lead to the intended overall result, but moreover, doing so more often than not has unintended and surprising consequences in places that you cannot see.

Translation companies have optimized their process and this has led to suboptimization. They have — logically — focused on maximizing their profit, not taking into account what this does to the overall value stream. So even though they often have the right intention, their entire operations are focused on efficiency of (their part of) the translation process, and not on the overall business goals of their clients.

The big problem that this situation causes is the fact that problems are hardly ever structurally solved. Even when there is a long lasting relationship between a translation agency and a client, and they claim they are working upstream to improve the process, there is usually no one person responsible for the entire process of multilingual content creation. Because this would take a process manager on the client side that systematically monitors the process of the translation company, or someone from the translation company do the same on the client side. Both very unlikely scenarios.

This means that structural changes to the content creation process, systems or client side translation processes in order to enhance the translation process on the side of the agency are very unlikely to take place.

In rare cases there is alignment of processes, but this only happens for the very large enterprise accounts. And even then, cases of real integration of processes, where content creation and translation are one, and true collaboration and feedback loops are facilitated are a needle in a hay stack.

Why is this so hard to change?

The narrative that most translation companies use in their marketing is that they solve your multilingual problems. Work with a translation agency, and you are relieved of all this tedious work that you encounter when publishing in multiple languages. This message has as a consequence that companies do not understand that a lot of the tedious work actually originates in their own processes and they need to take ownership in order to really solve the problems that are encountered.

Translation agencies are paid to “take care of things”, so in most cases, the client perceives them as the “problem-fixer”. It is incredibly hard for an agency that is in this position to influence the process on the client side. My personal experience while working for a translation agency, is that most clients are surprised if you ask questions about anything related to their business goals and what they need in order to achieve those. Although this should always be the starting point for any activity.

The same often goes for the localization managers that are running translations within a company. Even they are not able to influence the processes that really need improving, because translation is seen as something completely different from the core business of the company.

My hypothesis is that this is a direct consequence of the narrative of translation companies, that spread the vision that translation is something to outsource, that it is a nuisance that you do not want to have anything to do with. Obviously, they have to tell this story, or they would be out of business real soon. But the question is whether it is true, and if it doesn’t increase the issues with multilingual publishing instead of solving them.

This narrative is strengthened by the business model of translation companies. Because you pay for a translation by the word, this completely obscures the fact that the translation agency is doing everything but translating. They do project management, they source and train translators, they build tools and many other things. And this is all paid by the word. No wonder no-one seems to realize that a translation process entails much more, both on the side of the agency as internally, than translating words.

I believe another narrative is needed to solve the real problem. And that is: take ownership.

How to solve it?

What we see is that tooling and processes around translation have become extremely complex, exactly because of this waterfall approach. No wonder a lot of companies want to get rid of the tedious work that needs to happen when something that went wrong in the beginning of a process, but wasn’t too big a problem, explodes in your face at the end of a project when the severeness of the problem is multiplied by the number of languages you are translating into.

However, if companies take ownership of the translation process, (and refrain from treating their localization/translation department in a waterfall way), they will start to see that translation can point to a lot of issues in the creation of the content. Solving these can actually make that a more scalable process as well. Only then will the real cause of the problem be fixed, and this will enhance not only the translation process, but the entire process.

Once you start removing waste systematically, you will see that translation project management is no different than any other project management task. I claim that if you integrate your processes and apply an agile and/or lean approach, you do not need any experts in translation. You will get expertise about the several languages straight from your translators.

With some advice on the right tooling and processes, you will be able to build a smooth, automated process for all content, whether it is original or translated content, and this will benefit cost, speed and effectivity of your multilingual publications.

Should LSPs cease to exist?

You might think that I believe that Language Service Providers (aka translation agencies) should not exist at all. This not entirely the case. I believe that if they can reinvent themselves, and their business model, they can really make a difference when it comes to multilingual publishing. All they need to do is offer the services that are now all included in the word price as separate services and support their clients that want to take ownership. Examples are: sourcing and educating translators, paying translators, consultancy, solutions.

This will have an amazing impact on the translation industry, as a lot of superb solutions are now hidden from the customers, and are built only to make sure their client is locked-in on buying words. And who can blame them? As long as organizations are doing what they’ve always done, selling words is the most profitable thing to do in the translation industry.

What could happen if organizations take ownership of translations

If a company understands how to build processes and automate them, and takes the entire process into account, beautiful things can happen. Translations could be published at virtually the same time as the original, source and target content would be treated in the same way, the process would be easily scalable, meaning that adding more languages or more content is a no-brainer. Their content would improve because of the commitment of the translators, who are now included in the process instead of seen as a typing machine. And all waste could be removed from the process, leading to a decrease of costs, both internally and externally.

The quality of the product would be the same for all users, no matter the language they speak. Isn’t that what we are all trying to achieve?