Finding New Business Segments Through Big Data
By: Matthias Heyn (SDL) and Michael Wetzel (Coreon) - SDL plc
15 May 2015
LSPs should look at new opportunities to expand their services expansion in response to the widening impact of Big Data. Authors Michael Wetzel and Matthias Heyn contend that LSPs are in the right place at the right time — they only need the courage to step outside of their “comfort zone” and help clients break down obstacles caused by the new realm of Big Data.
Inside and Outside the Comfort Zone
LSPs are doing fine. The localization industry is growing nicely, year by year. But at what pace? Consider how dramatically the amount of digital content increases every day.
“Doing words” will certainly remain a comfortable revenue stream for LSPs for quite some time. While there may be no “Uber of Localization” yet, technology is still poised to gradually replace many of the activities the business model of LSPs is based on. Therefore it is wise to leave today’s comfort zone to explore new shores and develop new business opportunities. The demand is obvious and the challenge is discussed at every conference in the language industry: how can we cope with the globally exploding amount of content? How can companies can handle all the multilingual data thrown at them that needs to be mined and understood?
LSPs have unique expertise in global processes, in multilingualism, in key language technologies, and are experts in outsourcing. This knowledge can be leveraged for new tasks. LSPs have the opportunity to go after opportunities that will complement and maybe eventually outpace their current revenue streams. Acting early means conquering business and markets before the late-comers find the best land already taken.
Understand New Segments, New Opportunities
Imagine a customer support center of a global company. A product manager wants to be proactively alerted in a daily summary report about worldwide trends in incoming incidents, to react quickly and improve the quality of the product or service. Easy? Not really. Communication with users worldwide goes through a variety of channels in dozens of languages: chat transcriptions, dialog systems, keywords, auto-tagging, self-help tutorials, knowledge bases, incident reports, tweets, blogs, comments. Most of the data is text, predominantly unstructured. This huge amount of mostly “uncurated” content must be searched, categorized, mined, and analyzed, across languages and spanning varying levels of expertise and accuracy (typos, imprecise wording, or grammar deficiencies). To make this work, companies must deploy lots of multilingual technologies: text parsing and tokenization; terminologies, taxonomies, thesauri; translation memory and machine translation. Sound familiar?
Is machine translation the answer? Using MT as the only solution is a dangerous approach. Its quality depends on available resources in language and domain. Inaccuracies multiply along the process: 80% MT accuracy multiplied with 80% accuracy in sentiment analysis results in a hit rate of only 64%. At a few percent points lower you might as well flip a coin!
Summarization, auto-classification and categorization, text mining, social media monitoring, free-text survey evaluations, and question answering are all among the many recent data technologies that will soon be mainstream in enterprise processes. They all depend heavily on language technologies and the availability of domain specific multilingual resources — a core competency of LSPs and their ecosystem.
Identify Obstacles, Fill In Gaps
What is missing to get where we need to be? For us, a tool that proved helpful to start with is the so-called business model canvas. It allows you to describe, design, challenge, invent, and pivot business models. It forces you to nail down and summarize on a single white board what the value proposition is, identify required resources, what components to buy, which to build, outline main channels and customer segments. And, probably most important, to understand with which partner organizations you reach the audience.
At SDL, the strategy was to build, acquire, and integrate technologies (theirs and others) in order to create an exceptional customer experience platform. Customer experiences span a huge array of dimensions: marketing, sales, media management, campaign management, eCommerce, plus self-service support in technical documentation, customer analytics, journey analytics, and much more. The task required a “multi-disciplinary” or multi-technology approach.
SDL developed an integrated solution that enables companies to deliver great experiences across channels, devices and languages. By using other technologies as well as their core language technologies and services, SDL enabled users of the SDL platform to better reach, convert, and engage with their customers on a global scale.
Expansion through acquisition is not everyone’s strategy, of course. But however it is done, you have to identify the gaps to fill, your key resources, and the required complementary players (partners) to get the job done.
Experts in Multilingual Knowledge Management
As with every new field there are already quite a few companies out there tackling challenges in processing knowledge; knowledge contained but hidden in unstructured texts. But most technologies are available only in English and, if lucky , maybe for some “bigger” languages such as German, Spanish, or French. Others are excluded from the so-called data revolution. Making Knowledge Management global is where LSPs are in a brilliant starting position. LSPs own or have access to key resources. They can deploy and offer services around multilingual knowledge systems – derived from terminology databases or translation memories. Most importantly they know and talk to the people in large corporations who own Globalization. It is time to leave the comfort zone and start to discover new business segements.
Michael Wetzel is a co-founder and Managing Director of Coreon. Coreon is a pioneer in software for managing Multilingual Knowledge Systems – semantically enriched terminology databases that facilitate crosslingual search or multilingual text analytics to mine unstructured textual big data.
Matthias Heyn, VP Global Solutions at SDL knows about customer experience management, global information processes and translation production optimization and pharmaceutical regulated language processes. Matthias Heyn consults many large scale organizations for SDL and today is responsible for global solution design mainly for key accounts in the public sector and pharmaceutical industry.