April 2018
Text by Dave Ruane

Image: © rawpixel/123rf.com

Dave Ruane works on the enterprise business team at Xplanation. He has been in the localization industry since the mid-nineties and regularly contributes to industry events and forums. He is co-founder of the Process Innovation Challenge, a platform for innovation in localization. Current interests include continuous localization, global content marketing, innovation and simplification in localization.

Twitter: @DaveRuaneSpain 

The need for speed of global content

Within content enterprises, the continuous delivery of content in multiple languages has quickly become a basic need. How did we get here and how can we ensure meeting customer demands?

In 1998, software releases for a large desktop suite could be separated by up to 18 months, and even more for an operating system. Translated versions of the software could take even longer. Simultaneous shipping (Sim-ship) of different language versions was a pipe dream. At the time, translated versions of software could ship three to nine months after the English product – a gap that was commonly known as the "language delta".

Fast forward to the present day. Customer requirements have changed significantly and so has the ability to meet those demands. Audiences are used to receiving new content on an ongoing basis, and expect this. Companies like Hubspot and Marketo never stop producing content and constantly feed their audiences with updated and relevant information they can access through different channels. This "new normal" has driven investment and a focus both on customer experience (CX) as a way to propel brand interest, loyalty and revenue, as well as on continuous delivery (CD) as the mechanism to get products and content out there. In order to meet global demands, enterprises typically require continuous content delivery systems, which enable a predictable output of content 24-7.  

Continuous localization

Continuous delivery (CD) is a way to optimize the delivery process and to ensure that product and content deployment happen predictably and quickly. CD is based on Agile methods of content and product creation and is extended to the overall delivery chain. For CD to work, it needs to be applied systematically throughout the entire product or content life cycle. Continuous localization extends CD to global content while maintaining a seamless content workflow and user experience. 

Some of the outcomes of a functioning continuous localization process include:

  •  Improved time-to-market for global content and products
  • A "right-sized" output that ensures expected/predictable outcomes for customers
  • Improved productivity and efficiency in the content process
  • Improved customer satisfaction and a platform for creating improved global customer experiences
  • A reputation for the brand that is associated with speedy delivery of content and/or product

The software industry has led the way in terms of driving new technologies and trends and in many cases created customer expectations for continuous updates (think Microsoft Windows update as a platform and game changer from 2000 onwards). As software organizations took on Agile (The Agile Manifesto, 2001), which extended into Continuous Delivery and DevOps methodologies for developing products and content, the ability to produce reliable output in a continuous flow took shape.

Figure 1: Continuous localization is a subprocess of continuous delivery

Source: Dave Ruane

Continuous localization adds another layer of complexity to the agile content cycle, as shown in Figure 1. Often, it means live translations performed by humans. Take, for instance, the Facebook Safety Check. This social media feature makes it possible for people to mark themselves as “safe” during an event of crisis (e.g. a natural disaster or terror attack). People in the affected area need to be able to understand and know how to take action in their own language instantly; Facebook, on the other hand, needs to have the technology layer and on-call translation supply chain that makes this possible during such an event.

Evolution models and the constant of change

With his so-called Kano Model, Professor Noriaki Kano teaches us that "over time something considered a delightful innovation becomes another basic need". CD has come about due to the need to update products and content more regularly, thus generating a better customer experience. As of today, CD is a delightful innovation for some sectors, while for others, such as technology, retail, and travel technology, it has become a basic need. Revolutionary trends in content creation and digital marketing have made CD systems the "new normal" and, in effect, they demand that products and content are delivered in a continuous, speedy and predictable fashion.

Figure 2: The Kano Model 

Source: Wikipedia


The "content product" will continue to change and evolve in terms of volume and type of content, the speed it is provided at, and the platforms on which it is published. A mere five years ago, manufacturing companies commonly used translated online PDFs for providing technical support or product guidance. Today, they are providing the same information on multiple channels, be it Augmented Reality demos, Virtual Reality training, chatbots for customer support, mobile apps, etc. The same information that resided in 100-page documents has been transposed to be modularized, dynamic, and connected.  

A fundamental requirement for employing CD practices is to produce a considerably greater volume of content over a set period of time. So, how can you ensure a fast and predictable turnaround of content? Let’s start with the right technology.

Technological backbone and connected content

For their end-to-end digital infrastructures to maintain Intellectual Property (IP) requirements, organizations need localization cycles to be more efficient and secure than ever. As products and content have moved to the cloud, all content repositories (whether a Content Management System, Product Information Management, Product Lifecycle Management, Marketing Automation Platform, Software Configuration Management or other) will send content down the funnel in a fast, automatic, and consistent manner. The translation workflow has to pick up this content, process it, and deliver it back. All this must happen within an allocated sprint time. The approach requires newer technologies that connect to each other and have the ability to "level up" or transform as the "new normal" kicks in. While the Translation Management Systems (TMS) of today can cover some of these needs, the overriding feeling is that open connectivity has to be more standards-based to ensure it can cater to the growing platforms and volumes. One such current initiative is TAPICC (Translation API Class and Cases).

IBM, for example, has tried to stay ahead of the curve by building cloud services to enable the transformation, advocating APIs and SDKs as well as plugins and by deploying services that enable development teams to step back "and let translation happen". These interfaces are built into the DevOps toolchain so translation processes are triggered automatically. Development teams, if they wish, can enable transparent translation updates of their applications – so they don’t need to rebuild or redeploy. The key is a truly integrated development process where translation can happen without any disruption.

Customized and standard processes

Technology alone won’t do it, though. The translation processes must reflect the values that underpin DevOps and CD. A one-size-fits-all, monolithic approach to translation no longer works. Take translation quality testing, for example – some applications still require rigorous review, others not so much. When you can fix a minor bug in a matter of moments, does it make sense to slow down the delivery of the language version to make sure everything is perfect? Decisions around the process by product managers in consultation with their teams must be data-driven – data that is gathered by rapidly deploying, measuring and adjusting.

In an agile environment, scope can change quickly and frequently. This leaves no room for improvisation; instead, CD requires that rigorous process plans and backup plans are already in place. This requires good resource management to maintain flexibility.

If you have your content classified in terms of user visibility, audience, frequency of use, and the like, you can make faster decisions regarding the approach you are going to take to translation: the CAT (Computer Aided Translation) and MT (Machine Translation) tools you use, the amount and timing of any review or whether you translate at all. This should be reviewed on a regular basis to ensure future-proofing for next-generation technologies and continuous adaptation to changing customer requirements.

Asynchronous localization

Figure 3 shows the content lifecycle with the grey box indicating where the localization phase is currently integrated, and the blue box in the row below where it should be integrated.

Figure 3: Is localization moving to the left? 

Source: Dave Ruane


The more you can automate translation, the better. This includes automated localization testing, pseudo-translation testing, creating translatable user interfaces (which is particularly important if you are utilizing MT), using standard localization libraries, and the like. Building quality in from the beginning is critical when you are dealing with short sprint cycles.

The process should be simple enough that all a DevOps engineer or an agile content author has to do is press a button. APIs and plugins help you to create a pipeline flowing from the repository through the CAT tools to the application with human translators plugging in for review at any time in the cycle. Ideally, developers and authors are focusing on their core activities, while translation is happening asynchronously and almost invisibly in the background. By doing this, you are pushing localization further to the left of the content life cycle.


Continuous delivery is continuous adaptation

To spread global content, CD procedures across the content life cycle enable all stakeholders to work closely together. If we need to deliver a large amount of content in a short period of time, we have to step away from our traditional standard delivery terms and methods to make this happen. A flexible, yet reliable process will allow automation to be optimized, so the limited time that you have can be focused on the actual translation work. This includes modularizing content into smaller chunks, adding or modifying content on the fly, developing CMS connectors to automate translation ordering: the wheels have to keep spinning continuously.

What is global content quality?

In the age of CD, we have to broaden our views in defining quality. Quality needs to be impeccable for specific content and purposes. In other cases, prioritizing "urgency" over "total language and linguistic quality" is a trade-off required and motivated by business needs. Millennial companies have realized that users and consumers want immediate access to the right information. They are willing to adapt processes to meet publishing deadlines and articulate the quality levels required depending on content and audience. At today’s speed, translating “okay” can be good enough, in particular as tomorrow’s (or the next) release can bring any required fixes or translation edits. Many companies are testing these assumptions now in a way that was thought impossible a few years ago. They are "leveling up" from linguistic quality metrics and looking at the bigger customer experience picture; "if it doesn’t negatively affect the customer experience, then it is a non-issue". Linguistic issues, which in a previous release would have brought the delivery process to a halt, have now been reduced to "Priority 2" items in a continuous localization world.  

Together with content publishers, translation companies assess quality output requirements for each iterative sprint cycle. It is only by becoming part of the content life cycle that shared mutual success can happen. Another consideration is that in five to ten years’ time, the buyers will be people who have grown up in a mobile world in which typos and shortcuts are tolerated as long as the message comes across. The localization ecosystem needs to get on board with the new quality paradigms that are driven by tomorrow’s buyers.

Holistic approaches

Technology alone, as we saw, won’t enable real-time continuous localization of content. A holistic approach is needed, involving a content development program that is modular and has a single-sourced content-sharing vision. Such approach can support next-generation platforms and content distribution technologies, including AR/VR and bots.

Strategies that work include a program of modularization topics, using smart platform-specific markup to enable responsive content, and single-sourcing all content so various delivery platforms can get a unique, rich, on-demand, consistent experience.

Smart workflows enable technical writers to get complete and translated modules in the CMS quickly and easily. On the global content side, they manage tasks like quality control of the linguistic process in a continuous delivery flow. Key to keep in mind is an Agile mentality and a culture of change, knowing that as content delivery platforms are shifting and changing, so should their approach. Short delivery cycles ensure that content can be adapted to new platforms and technologies such as bots for customer support and Augmented Reality for sales demos and training.



Some considerations when embarking on a continuous localization journey:

  • Do you have continuous localization in place for the full content spectrum? You might have some core content built using Agile methods, but do you also use it for the technical guides and support material? Is your marketing team using Agile in their MAP system? Do you need to align various content teams to a common sprint schedule?
  • Technology: How old is your current workflow technology and stack? What does it not support?
  • Modularization: How low can you go with your content topics and still maintain good context?
  • Maturity model: How mature is your Continuous Localization model? Is it tactical, measured or optimized?
  • How simple is your Continuous Localization model? Do multiple content repositories need a separate translation workflow? Do you still copy and paste some content to send to translation?   
  • What connections exist between content, development and localization teams? Do they need to move closer together in terms of goals and a shared culture to ensure the overall success of the global product?



The future is now

Much more and smarter content is around the corner. Enterprises will need strong and tuned continuous localization machines to cope with this growth – which will include new sources, formats, and language requirements. Content will also get smarter (think chatbots and audio assistants), and the role and use of content will adapt. A strong, future-proof continuous localization methodology means moving from something that is a "delightful innovation" to it becoming a basic need. Are you ahead of this curve?



Further Reading: