14. Gengo: Winning in Translation Through Crowdsourcing
Who else would understand better the value of enhancing intercultural exchange than Robert Laing and Matthew Romaine, the founders of Gengo? Both of them are multicultural. Robert was born in Australia, and has lived in Britain, Belgium, and Japan. Matthew has American and Japanese lineage and has also lived in both countries. Both of them have experience working in intercultural projects. Matthew had worked with Sony as a member of its international growth strategy team. He was asked to translate a lot during his work there, up until he decided that he wanted to get back to his roots and experience working in a Japanese company. Robert, meanwhile, had previously supervised the creative team of a global branding agency. He wanted to build a company matching his global experience. Both of them got together and created a company (initially a “side project”) which offered a human-powered platform for natural, native sounding translations. The company Gengo1 (meaning “language” in Japanese) was born in June 2009,2 making the world an even smaller place.
1Gengo was initially called “Mygengo,” as the domain name “Gengo” was not available then. However, when they later managed to get the domain name, they rebranded the company as “Gengo.”
2The service was launched in December 2008, before the company was established in 2009.
Gengo is a Japan-based company which offers translation services, allowing people to communicate freely across any language. These services span around categories like e-commerce, CMS, Internet media, etc. Gengo employees include about 15,000 translators across 114 countries. It handles both word-based and character-based languages. Currently, there are 37 languages and 63 language pairs that Gengo handles and the list is still growing. It is planning to include three more pairs to this package—English to Serbian, English to Slovak, and English to Ukrainian.
WOW: Crowd-Based Translation Services
The prices are consistent irrespective of the language pair chosen. However, as the quality goes up, the charges also go up. Some of the clients of Gengo are Alibaba, LinkedIn, Mozilla, Airbnb, Huffington Post, TripAdvisor, and Shiseido. Gengo is also receiving a huge volume of jobs from the travel and e-commerce industries. The company has also grown through collaborations with partners like YouTube and 3Play Media. Gengo is one of the two integrated paid translation services of YouTube.
By 2014, Gengo had received a total of $18.8 million. Investors included Dave McClure (500 Startups), Felix Miller (last.fm), Joshua Schachter (Delicious), Brian Nelson (ValueCommerce), Christoph Janz (Pageflakes), and Benjamin Joffe (Plus Eight Star). Gengo is registered as a company in Delaware, but has its headquarters in Tokyo, with an office in San Mateo, California. San Mateo office has about 10 employees (designers, sales, and marketing staff).
Strategies
How Does Gengo Translate Their Vision into Reality?
Gengo is a pioneer in the translation industry, capitalizing in crowd management and technology. As Robert notices, like in every industry, incumbents in translation industry are engaged in working on high-quality, high-profitable models. Gengo, started with a low-profit model, and then rapidly increased its quality, capacity, and profitability. The quick, scalable, and integrated services Gengo provides have earned it a huge portion of the $33 billion translation market. As on May 2014, it has translated over 200 million words. The translation rate is four words per second (i.e., 240 words per minute). In Maitri Jani’s (ex-PR Manager, Gengo) opinion, in a minute Gengo can translate at twice the speed of a live, verbal translator and at six times the speed of an average person.
Gengo noticed that most of the traffic happening in the platform, a.k.a. 75% of the jobs, were for translations under a 500 word count. Most of these contents were also dynamic in nature. This lead to the launch of Gengo’s API in April 2010. With the help of API, developers can integrate Gengo’s translation platform into websites or other applications. API thus helps manage the large changes, which are required in translating dynamic content (e.g., blog posts). YouTube is one of the main users of Gengo’s API.
OOMPH: Gengo as a Recruiter
Gengo receives as many as 8000 applications every day from aspiring translators around the globe. The passing rate to become a translator is approximately 7%. Based on the performance in a two-stage testing process (machine-tested multiple-choice test followed by human-reviewed test), the first level of translators, a.k.a. “Lay Translators,” are selected. About 15,000+ Lay Translators are selected from the applicant pool. To climb up successive notches of the ladder (Business and Ultra/Proofreading levels), the translators have to pass additional tests. Higher up in the crowd translation pyramid, the Senior Translators develop and evaluate the pool of translators, oversee testing, and assure quality in their respective language pair. Their other responsibilities include conducting performance reviews to assess translator performance, ensuring timely turnaround of jobs, and creating language-specific translator resources. Professional translators with good experience and track record could apply for this position. About 65 Senior Translators work at Gengo now. Further up, on the apex of the pyramid, the internal “Tops” (Translation Operators) Team consists of four members, who oversee the performance of Senior Translators.
Gengo does not do targeted recruiting. The potential employee pool does not focus on any specialist domains. If the customers require expert knowledge for certain translation tasks, Gengo collaborates with specialist companies to carry them out. The quality of the work is made available individually, to each translator. However, after noticing that some translators are broadcasting their scores publicly, Gengo is planning to publicize the scores of all employees from their end itself.
Gengo as an Open Data Source
Traditionally, translation industry is quite subjective. However, as a people-powered translation platform providing services based on data, Gengo places great importance on transparency, which is currently lacking in the translation industry. The company has an Open Data Initiative, under which Gengo website gives information about the number of words translated in real time. Gengo’s metrics from the measurement of translation quality to the assessment of support performance is available publically (and is updated every week!). Real information on every aspect of the services including support, speed, capacity, and overall performance is available on the platform. This way, the customers can make informed decisions. Gengo feels that providing persistent, complete, and accurate information is a beneficial thing for the company too. They consider this as a way of ensuring them being honest not only with the customers, but with themselves as well.
Gengo as Symbiotic Company
The whole principle behind Gengo is about giving back to the local society. Gengo has connections with local reporters (TechCrunch, Financial Times, etc.), embassy, and local media reporters. This has ensured them some recognition and market visibility. By partnering with and by supporting local entrepreneurs, Gengo advances its symbiotic presence in the local Japanese market. Another way they do this is by collaborating with interested intrapreneurs in enterprises. Gengo also makes sure that it hires local talent more.
Gengo’s Quality Policy
Gengo has a unique quality policy with a metric-based assessment mechanism for examining the quality of translation services. The quality is influenced by factors like the understandability of text, clarity of instructions, context, etc. Gengo catalogues the quality differences offered by its different translation levels, a.k.a. standard, business, and ultra. Standard translation offers largely accurate output and a 7.0+ Gengo quality score (error tolerance). Usually part-time translators with high proficiency in the language pair take care of the jobs. At business level, the translation is for professional usage. It has a quality score of 8.0+ and is taken care by translators with advanced or professional translation skills. Professional use translations can be accompanied by additional proofreading under the ultra option. Here, the translators have advanced/professional translation capability with proofreading skills. A quality score of 9.0+ is provided at this level.
SO WHAT Is Gengo Translating Next?
Gengo provides accurate and natural translation services, when compared to a machine translator. Matthew Romaine thinks that machine translators are not competitors, but rather complementary service providers, when compared to Gengo. In his words, language is a living organism. There would always be a need for human translation. Nevertheless, Gengo could offer machine translations as additional, first level services.
Traditional translation agencies typically focus on small jobs, but usually cannot operate on bulk amounts of data. Gengo upgraded the translation industry by introducing a new technology and combining it with crowdsourcing. It provides translation services at a scale. By doing systematic testing, quality checking, and peer reviews, high quality of these services is also assured. All this is done at a much lower cost than a traditional agency, as crowdsourcing reduces overhead and unnecessary transactions. What’s more, the online business approach solves the problems related to distribution and geographical barriers.
When compared to translation agencies which have an online presence, Gengo punches above its weight by offering API. Gengo API simplifies human translation by making it efficient and reliable, just like a machine translation. The jobs are ordered, tracked, and managed right from the CMS. Up to 1000 parallel jobs can be handled at the same time through this.
The use of crowdsourcing and API also provides competitive advantages, like more convenience and reliability to customers. This is accomplished by reducing the number of steps and errors involved in the workflow of the company. The average response and delivery times are reduced as the translator pool is spread over all time zones. This results in comparatively shorter lead-time. The flexibility and scalability of Gengo ensure that the customer gets the best deal always. By offering a large pool of translators, regular customers can choose their “preferred translators.” This way the end result is consistent and in the required style. To ensure that their daily online postings have a constant tone, industries like media also usually prefer this option.
Low cost is another advantage offered by Gengo. The translators are of the highest quality. Traditional translators who offer high-quality services charge customers a very high rate. By specializing in simple, short texts rather than complex, long ones, Gengo makes the customer pay for just the services he needs. This reduces the cost involved. Gengo services are useful for most applications. By offering customers the option to reject a translation and demand a full refund, Gengo puts its money where its mouth is.
Compared to adjacent companies like Smartling, which offers translation project management tools, Gengo is uniquely positioned. Its strength is not just in its technology, but in its crowdsourcing model. The technology is suitable for high-volume transactions. But at the same time, the crowdsourcing component allows for reduced prices.
Challenges
Principally, the translation industry is a very conservative/traditional/non-technological industry. Apart from the resistance faced due to this, the lack of employee mobility in Japan also caused some sourcing challenges. The lack of flexibility in contracts was also another issue. Another main challenge Gengo faced initially was its low cost. Customers were initially skeptical about using cheap services for jobs which require high quality. However, by looking at Gengo’s trajectory, it is safe to conclude that customers have got over this notion.
Another thing that goes against Gengo, when compared to its counterparts, is that it appears to be providing more general services than the specialist services provided by competitors. OnehourTranslation, for example, caters to the same customers of traditional agencies, plus developers and digital marketing managers. In this way they are similar to Gengo. The difference lies in the fact that OnehourTranslation distinguishes between market segments, and have dedicated departments for each type of customers.
A related challenge for Gengo is the increasing number of competitors entering the market, with some of them providing a wider variety of services than Gengo. Crowdin, Verbalizeit, WOVN.io, etc., are examples. WOVN.io is a multilingual localization service offering both machine translation and human translation. Crowdin is a localization management platform for translation services. Transifex is both a translation and localization platform.
Conyac, a Tokyo-based competitor of Gengo, differentiates itself by having a tiered community review process. The finished work is reviewed several times for accuracy by other translators before being delivered to customers. Gengo responds to this comment by saying that their aim is more to domesticate translation (to provide reasonable, fast, and high-quality translation to people). Possessing API, which Conyac doesn’t have, gives them a competitive advantage here. Furthermore, they are rather happy that there are more translation options for people, so translation becomes more convenient for people.
As a final point, the heterogeneity of translation requirements and the associated data security issues raise the criticism that Gengo will never replace traditional translation agencies and Language Service Providers (LSPs). The fact that big LSPs have not yet turned their focus on this segment also arouses suspicion.
Outlook
From the time of founding (when it focused on just one language pairing), till date, Gengo has shown an aptitude to evolve and adapt. Gengo entered an archaic market, contended retention issues, large customer company naïveté, and unwritten employment traditions, and yet managed to be successful on a global scale. Even with an increasing number of competitors entering the market, this is what sets Gengo apart. It is the success story of innovation, crowdsourcing, and human acumen. Referring to Gengo as (better than) the “Mechanical Turk for translations” is not just a play of words (pun intended).
From the Perspective of Nozomi Umenai, Marketing Manager, Gengo
Here are answers (in verbatim) to our questions by the Gengo’s Marketing Manager. Even if in a brief format, they provide additional useful background on the company.
1. The purpose on which Company was founded; what is your passion?
• Gengo’s website was launched in December 2008, but the company was established in June 2009.
• Gengo was originally a side project for Rob and Matt, but they received needs soon after they launch Gengo website.
• Our passion is to create a world without language barrier, “communicate freely.”
2. Future intent; what is the next opportunity and related challenges?
• Continue to grow as a global company, help the world communicate.
• We will continue to improve our platform so we can provide the best translation experience for our translators.
• Will start to handle slightly more complex content and higher quality so we can cover more translation orders.
• Will open European office to handle Europe demand.
3. Company data, stats, etc.
• Established in June 2009.
• About 30 employees in Tokyo, about 45 globally.
• Office in Tokyo and San Mateo.
• Rakuten, TripAdvisor, YouTube, Alibaba, Quizup.
• Sales rep covering Beijing, Shanghai, and Europe.
• About 15,000 translators around the world.
• About 307,000,000 words translated.
• 37 languages.
• 7 languages pair from Japanese (English, Spanish, Korean, Indonesian, Mandarin, Cantonese, and Thai). Thai is the new edition launched on August 5, 2014.
• Only 7% of the translators pass the test.
• Translation done in 8 hours in average.
• 95% of translation picked up in 60 minutes.
• About 1000 English to Japanese translators (and vice versa).