
Localization 3.0, the Solution for Transformation
By Anson Zhang
Localization 3.0 is the next era for delivering localized products and services including mobile apps, SaaS applications and web contents in particular to global customers, linguistically and culturally. It is not just a new model and process but also a disruptive mindset for the industry to get adapted to. It brings fundamental changes for various role play including development, release management, localization management, translators and end users, yet creating a better world where we are offering and accessing new services and experience.
Why Localization 3.0
Tech world is evolving and advancing faster than ever before. Looking at the industry of localization today, it demands “out of box” strategy and cutting-edge ideas to respond to transformation and changes.
Localization 3.0 is that response. Back to 80s when localization started to play its role in product globalization, over two decades or so since then, representative offline desktop tools Trados Translator’s Workbench and PASS Possolo (later acquired by SDL in 2005 and 2007) were the dominant productivity tools for documentation and software UI localization. Translation was either done in-house or outsourced to global vendors but fully offline. At this early stage, localization history was heavily manual and labor intensive, making an initial age of Localization 1.0 (L10n 1.0).

Moving to late 90s till the first decade of 21st century, online translation management system (TMS) emerged like powerful Idiom WorldServer (acquired by SDL in 2008) as a typical TMS to enable translation agencies and large tech companies to deploy on internal server to manage translation online with limited integration with source management systems while translation was still largely completed offline using desktop tools. Localization 2.0 (L10n 2.0) witnessed such model until around 2011.
With the booming cloud and mobile technology over these years, conventional localization technology giants are all legging much behind to keep up with the disruptive tech world and do not fit well into the changes in localizing mobile and could applications in particular. Shifting from waterfall to agile development also demands speed for a continuous delivery and hence calling for fundamentally different localization model, process and tools. The solution for this change is Localization 3.0 (L10n 3.0) and it is on the air.
L10n 3.0 Model
Pull and Push
Automation is not an option and it is a beauty. Traditional L10n 1.0 and 2.0 technology and process involves time-consuming and manual steps for example preparing translation kit, sending it to vendors or creating a translation request on Translation Management System (TMS) and translators working offline without any visibility which makes it hard to track real time progress and difficult to manage schedule for today’s fast-pace delivery request. It also consumes tremendous time to drive multiple-layer offline communications and collaborations across time zones. However, changes started since around 2010 when a few of localization technology startups that are building cloud-based and continuous translation platforms such as Crowdin, OneSky, AppLanga, Smartling, SmartCAT, Mojito and quite a number of others. Some of these solutions are designed particularly for localizing mobile, web, game and cloud applications.

* Click here to view an interactive version of this graphic hosted on Tableau Public
Unlike traditional L10n tech giants, these new technology companies interrupt localization model that has been the trend today and a promising foundation for the future as well for localization industry because traditional TMS no longer fit into the new process. It requires all internal systems are fully integrated to auto pull source files or strings from code repository or from within application for translation online and then push back translated target resources for continuous build generation or even directly to end users basing on over-the-air (OTA) technology of new releasing model. Conventional technologies in particular are lagging behind from this new change just like Microsoft lost its mobile market share and online payment was not started by traditional banks worldwide.
Another major urgency for localization under the new model is to move all activities online including translation and collaborations in particular. Only bringing production online can fundamentally streamline the whole localization lifecycle and leverage much more potential benefits of new technology, cost control and quality efficiency. What’s more pressing is that a continuous translation process is mandatory for today’s product deployment lifecycle under SaaS business. Conventional localization technology provides very limited capability to support these changes which have been inevitable and it demands either make or buy. Some companies either large or smaller like Microsoft, Box or Evernote, they build and use their own customized localization tool chain while most of other companies may have to buy and then customize bit to fit into own specific scenarios to reduce overhead so that localization management is able to focus on real high value activities.
Connected
Offline process is no longer working or needed. Localization 3.0 is about innovation on localization enablement, full automated process empowered on continuous localization platform but also introduces new localization business model in which everybody and everything is connected. This model has bottom-up features and some examples are:
- Transparency,
- New engagement,
- Fundamental role change for translation vendors,
- A global talent alliance to be built up as an eco-community for providing linguistic quality as experience,
- Relying more on product experts to translate instead of linguistic professionals and,
- Certifying translators to build exclusive and predicable language capability.
- Quality traceable with incentive plan, for example, translators will be rewarded with bonus for zero linguistic complaint within first 2 years since the release, etc.

Localization project managers at upstream in particular will adapt to new and expanded responsibilities, for example from traditional role of planning and executing localization to driving localization design for each product or new feature, from project management to program management, from working in project to working more on data, service and customer experience because L10n 3.0 is not only changing localization team, translators but also development team as it is a different way, creative way yet much better way for tech companies and end users to provide and access localized services.
Experience
Experience is everywhere. Product development, localization management, enablement engineering, end users and translators are all the stakeholders in this new model. Localization as service is the key in the era of L10n 3.0 that simplifies experience for all parties through adapting new and unified framework for all of your products to innovate localization at coding level or using fully integrated process to improve speed and engagement to reduce touch points and eliminate paint points.

L10n 3.0 also reaches out beyond delivering localized interface of products to ultimately end users and customers, for that, delivering localized experience, in a cultural sense in particular, will be the main focus as well because all things that human is doing is for human, so does localization. On process level, it makes everybody’s life easier; in business prospective, it will be a valuable subject to truly create service local for local to improve localized user experience.
Given less yet paid more. Although Machine Translation quality is improving over recent years thanks to the changing machine learning technology, MT become increasingly smart as a result, which enables higher efficiency and less efforts from translators. This will be particular true in IT localization, but L10n 3.0 does not recommend decreasing rates of translation but gradually increasing the rate to stay focus on human to deliver “WOW” experience to delight human. Quality is free or costly, never cheap.
L10n 3.0 Process and Delivery
In L10n 3.0, source files or strings are continuously or regularly pulled onto the continuous localization platform and translation becomes on-going effort which means localization is an iterative life cycle at the speed of feature development. Resources from master branch and feature branches are being dynamically pulled and translated to enable a maximum sim-ship with source language. Post translation updates are directly pushed back to next integration or to application end users real time over OTA technology as independent translation update release, for example SDK embedded in application upon connection will auto call cloud-based string distribution server to fetch latest translations or even new target languages added available on the server.

L10n 3.0 model enables translators to easily communicate online and consult each other to leverage support and experience, and translators also have instant assistance from client, project managers and product linguistic experts to resolve any query or challenges quickly to simplify and speed up translation. Various productivity plugins are installed and reporting features are available for self service. Besides, centralized linguistic assets such as TM, glossary and legacy data are all available online within single portal to reduce touch points so that all teams have more time to focus on key activities.
Traditional screenshot taking and review which is a time consuming linguistic test process is no longer needed in L10n 3.0 model because translation will be done in context for mobile app, cloud application and web contents in particular.

Rich context can be provided to translators in L10n 3.0 model, for example string contextual comments from developer or project stakeholder, actual UI screenshots taken automatically via SDK and tagged automatically basing on string ID or key, or even virtual in-browser interface of the applications being translated. All these references are visible within the online editor to translators, hence making sure of delivering right the first time to dramatically shorten schedule and minimize 3rd party review for time and cost savings. For web contents, translators will work in live content to have actual preview on translating, and even if working on html files, it is all a WYSIWYG process.
With that, linguistic quality is guaranteed much easier than ever before. Linguistic testing and translation are simultaneously done as one step or even no separate linguistic testing is needed again after translation is done, which improves efficiency fundamentally with truly faster Time-to-Market. L10n 3.0 is more a disruption than changes or improvements in terms of model, process and technology.
L10n 3.0 Solution Architecture Overview
Following graphic shows a general high-level infrastructure of L10n 3.0. It’s a mix architecture to support various major scenarios basing on user story at different company — file based, string based or hybrid model. Some companies may adapt a much more cutting-edge architecture in which resource are pulled and pushed through a unified middle-layer service directly communicating between application and translation platform to bypass integrating code repository.

Leveraging cloud, SDK and APIs to enable translation at the speed of continuous deployment of for example mobile apps and SaaS applications is the goal of L10n 3.0 model. It provides productivity, flexibility (adding new language or pushing translation updates in real time without delay to end users by decoupling with build process,etc), consistency, cost efficiency and better all experience.
How to do it? The first thing to adapting L10n 3.0 are to auto pull/push linguistic resources for continuous integration or disruptively to transform product code using a unified internationalization framework across products to enable dynamic OTA delivery, and then automate process pain points and minimize touch points by setting up a centralized platform to connect all parties and all things online, and finally leverage data technology for business management and implement evolving intelligent MT to gain higher productivity and cost saving to architect a robust and scalable solution for decades to come.
Localization 3.0 has come. It integrates horizontally, connects vertically, focuses on experience and leverages power of technology. It is changing a lot and also creating a lot for now and for future. Evolving localization surely makes more feasible for the industry and global customers.
About the author
Anson Zhang is a globalization program manager at VMware Inc. He has more than 8 years experience in localization delivery and technology innovation. He is reachable at anson.zfy@outlook.com and Linkedin page at www.linkedin.com/in/ansonz.
