Life in the localization business is like one long obsession. We collectively believe it is important that the world economy recognizes our language business. But we cannot make the business really work. The localization industry has been developing for some twenty years now. But if you look back, you wonder whether it has been a success story. Even today, after much effort and experimentation, our rather small industry is in a deplorable state. What is wrong? Does language really matter after all? Are we all barking up the wrong tree?

While the localization industry is stumbling its way through these mounting issues, continually trying out new strategies, the real solutions to corporate needs for multilingual services are emerging from other corners of the business world. All the signs of change are there, but the localization industry does not seem to be paying attention.

User-centric information

The answers, although obvious to those who keep an eye open on developments, are likely to cast a pall of disillusion and dismay over many of the leaders in our industry. The business models built around MLVs (multi language vendors) and GMS (globalization management systems) will be even more acutely challenged, and may even need to be abandoned. If the language business is really important to us, then we'd better reinvent ourselves fast, or end up a mere fading memory of the late 20th century.

Language professionals should know better than most people that our relationship with information has changed drastically over the last ten years. Just think of the new verbs that refer to what we do with information: scanning, browsing, surfing, chatting, emailing, gisting. We do not read the manuals that come with the software we bought. We pull out the information when we need it. Which means that publishing quality is no longer the most vital issue. Instant access is even more important. Information must be consistent and current, and that prevails over style.

It is no surprise that IT companies over the past ten years have shifted their emphasis from printed manuals to online help and now to web-based customer support. The user is in control. His or her demand for instant access turns the availability of an effective knowledge base into an asset, and the lack of it into a liability.

Slow and costly

Against this background it is obvious that the current localization model is too slow and too costly. IT customers have come to depend on multilingual localization vendors as a necessary evil, because they did not want to increase their internal headcount for non-core activities like localization management. As a result Multilingual Vendor (MLV) companies emerged. And to secure their position and further increase their market share most MLVs invested in proprietary technologies like workflow automation.

However, the actual outcome of this strategy has been the exact opposite of what was intended. MLV revenues stagnated. The technology proposition, intended to improve the value and efficiency of the services, is often perceived by the customer as expensive and a trick to lock in his services. The markup on translation costs can be 100% to 200% due to multiple layers of coordination, QA and translation memory management. There is no evidence yet that technology can help to significantly reduce this cost.

In terms of total volume of throughput, the professional localization industry is already dwarfed by fully automated translation productivity. SYSTRAN alone translates 6 million web pages per day. Assuming for a moment that the ABI market sizing is more or less accurate, the total throughput of the human translation market amounts to approximately 550,000 pages per day [1], or less than 10% of the throughput of SYSTRAN.

Facing the facts

Even though the SYSTRAN Translate button is mostly a free service offered through the portals, this is a clear sign that there are viable alternatives to mainstream translation. Like Autodesk (who presented in Chicago in August 2001), more and more companies will shift to fully automated translation technology. The following factors will support this development:

• Call centers shift more and more to web self-service.
• Documentation shifts to online help and further to online customer support databases.
• Customer support databases accumulate massive content, sometimes 100,000's of pages.
• 100,000's of new entries are added everyday to various e-commerce sites.
• Human translation is simply not an option: too slow and too costly and not scalable.
• With proper customization a good server-based MT system generates satisfactory output, and most important real-time and on demand

The signs are very clear. The localization industry as we know it today will be marginalized unless we adapt to the new customs of "information consumption" and embrace the new technologies.

Globalization Management Software (GMS) offers few real advantages to enterprise customers. Content management software will take over the functionality of recycling multilingual content. Translation memory functions will be targeted at a smaller level of granularity: phrases and terms. Enterprise customers would much rather see some of the useful features of a GMS integrated in their content management software, than to undertake an additional expensive software installation for very few benefits. Localization companies, another likely customer profile for GMS providers, reject the software because of the competitive threat. The best strategy for GMS vendors will be to merge with a content management software provider.

Large Multilingual Vendors (MLVs) are seeing their technology investments failing, because corporate clients do not like to be locked in to one service provider. Large customers can find the same (or better) localization quality and service reliability at the smaller localization companies. Small scale localization companies generally deliver more consistent quality at lower costs. As a result the MLV continues to experience price pressure.

The collaborative future

This pressure will only increase with the arrival of new technologies and new approaches to workflow, that will allow many small vendors to collaborate in an open web-based infrastructure. Collaborative Product Commerce (CPC) and web services enabled by the UDDI (Universal Description, Discovery and Integration protocol) fit very well into a multi-vendor localization model.

The best strategy for a MLV will be to transform itself from a localization service provider into a consultancy, and act to implement translation and multilingual content technologies and open collaborative marketplaces. The bundling of proprietary technology and services is doomed to fail. Large MLVs will have to deal with a profound transformation of their business, and this will demand a critical change in strategy.

And finally, fully automated translation is set to thrive. There will almost certainly be a shortage of professional linguists to help with the coding of dictionaries and the development of customized systems. And this may well open up another opportunity gulch where MLVs can step in and help customers to implement the kinds of machine translation solutions which will gradually move in to center stage.

Jaap van der Meer
jd.vandermeer@quicknet.nl

[1] ABI published market size figures for localization and translation in the range of $10 billion. The localization industry represented by LISA is only a fragment, perhaps 10%, of this total. For this comparison I use an average cost per page of $50.