Bringing APIs to the translation industry

The first time I realized that application programming interfaces (APIs) were going to be everywhere was in 2006 in Barcelona, when I met the founders of 3scale, currently one of the leaders of API marketplaces. Every time I hear about how the translation industry is evolving into a technology sector, I think about APIs. In simple terms, APIs allow your software to connect to other IT systems, and be connected to third party applications.

How many translation companies in our market have a clear and defined strategy for APIs? The answer is not many, despite the fact that APIs are nothing new. APIs have already allowed the interconnection of business-to-business systems for several decades. In order to move technology forward in our $35 billion-plus industry, APIs should be part of translation companies’ offerings. The CEOs of translation companies will need to align APIs with their growth strategies.

There are countless examples of API usage outside the translation industry. Fully 90% of Expedia’s revenue is generated through APIs. In the travel sector, APIs have been extensively used, and are the core within travel technology leader Amadeus, since travel agencies and other stakeholders in the industry can use Amadeus’ platform within their own systems.

eBay and Amazon have opened up their platforms to allow third party developers to use the features within their marketplaces, not only to publish and sell, but also to create innovative services on top of that. Particularly for eBay, 60% of sales come from third party apps.

Social networking sites such as Facebook and Twitter have developer programs, with services that can be accessed through external applications. In the early days of Twitter, the company had a very powerful API that allowed the creation of third-party apps that boosted the usage of its 140-character service. Therefore, API strategy was a crucial factor for the company’s growth.

APIs are also a way to add modularity to software design. Evernote’s usage of APIs is more than 99% internal. With this, they achieved a flexible system architecture designed for growth.

In our industry, the objectives of an API for translation buyers might include:

 Adding the best providers for every service. Companies require a wide variety of services, and might need to outsource some of them that are not core to their business model. APIs allow them to integrate any third-party software into the proprietary platform in an easy way. For instance, you might want to integrate machine translation services from several providers depending on the language pair and field of a given task.

 Reducing costs and turnaround times. Most bottlenecks in company processes are related to human involvement, so minimizing the number of times people are involved during the translation process will reduce the company costs.

 Creating new business opportunities. Using APIs enables creative thinking, and might generate new business models thanks to a better identification and fulfillment of the customers’ needs.

Industry pioneers

If your company is a translation buyer, having a provider with API capabilities will improve automation and reduce the amount of manual work on your side. The more tasks become automated, the better it will be for your company bottom line. The benefits are clear from the financial side, but also at the process level, where your employees will focus on the tasks that add value and not on those introducing waste — you will be making your company lean.

The three largest translation companies in our industry — Lionbridge, TransPerfect/ and SDL — all have APIs, maybe due to the fact that changing the value proposition from services into technology can make any translation business much more profitable and valuable. In this sense, Lionbridge’s onDemand API business is growing at a rate of more than 100%, trying to automate every step in the translation workflow. TransPerfect provides valuable solutions to sophisticated customers with their suite of GlobalLink connectors and services, and SDL offers Language Cloud and has opened a marketplace so that developers can create innovative plugins for their products. Moravia and Welocalize also have developed good strategies aligned with people and processes.

Besides the big players, was one of the pioneers at the beginning of the twenty-first century, and there are other smaller companies such as Straker Translations, Lingo24 and Nova Language Services that incorporate technology as one of their core assets to maximize their value to customers. They are among the early adopters, but since most computer-aided translation (CAT) tools and enterprise resource planning (ERP) software products in our industry have made APIs available to third parties, developing this strategy is more a matter of determination rather than a project where your company might invest too much money without seeing results.

There are generic benefits from APIs and the continuous innovation process that should be realized at every translation company. Benefits of even more value include:

 Broadening the customer base and enabling new distribution channels. Related to the long tail of products and services, creating an API and succeeding in its marketing will allow a company to expand its customer base and gain traction from companies that would not otherwise use translation services. Different partners that have access to your API might be a catalyst for business model innovation.

 Increasing the number of services. Translation companies have experience with languages, and natural language processing is an extension of that. Technologies such as summarization, sentiment analysis, keyword extraction, recognition of entities and similar language technologies might enrich the experience customers would like to have.

 Increasing the value of the data. In this world where an abundance of data is created every day, translation companies own huge amounts of good quality data that can help in the creation of new services. We might see the translation company as a platform for language, rather than as a services company.

An additional benefit from APIs is that they allow companies to innovate with business models in the search to eliminate the price per word or the price per hour. One of the main benefits is that you can charge a monthly price for the availability of the service — integration, support and maintenance — and then add a price based on the unit the client wants. Perhaps that would be per character or per call, or maybe it would involve tiered or freemium business models. With developers using your platform, you can play with different revenue schemes. And it is essential to try to think outside the box and not blindly follow what the competition is doing.

A raw MT example

This year, my company was approached by a language service provider (LSP) that wanted to apply machine translation for a very specific content type for a software company client. The process the LSP had in place was the following:

1 Once per month, the client exported all its text strings from a database, and created a comma-separated values (CSV) file.

2The client uploaded the file into a File Transfer Protocol (FTP) server, and the translation company downloaded it.

3 The translation company created a project in a CAT tool, machine translated the CSV file, then downloaded it in order to upload it to an FTP server.

4 The client downloaded the file from the FTP server and imported the translated content into their database.

As you can see, there are many processes in the previous description that introduce waste, and our proposal was to try to minimize the human work, allowing the end customer to use an API to send the text strings whenever they were inserted in the database. Not only would we be able to provide real time MT instead of just providing the translated file once per month, but we would also minimize the human involvement. The only process we would carry out monthly was the statistical analysis of the most frequent structures in the source texts, which allowed the improvement of the MT output based on the linguists’ feedback.

After implementing this enhancement, our customer’s business model evolved from a price per machine-translated word to a better price definition based on what the client really wanted.

A human translation example

Proxy solutions have been integrating the automatic detection and translation of the content for several years. This approach can also be applied for any type of content and different interfaces. The first step for this is a client translation portal that many companies already have, where customers themselves upload the content and it starts the translation workflow we define, which can be fully automated. We can even go one step further by installing a robot that scans predefined directories in an intranet, and posts those files into the translation company’s ERP system so they can be translated.

We implemented this kind of project for a huge telecommunications company in conjunction with an LSP. We were scanning specified folders in real time, and whenever the robot found a new file, it was posted into the translation company’s proprietary ERP, and the translation workflow was initiated. The file to translate was converted to an XLIFF format and pretranslated. This file was automatically sent for translation thanks to an email containing a link to an online page used to translate the file. After the translation was delivered, the client added a quality assurance process before the translated file was generated and copied into the translated folder within the company intranet.

The human involvement for project management tasks was reduced to a minimum, and translations were successfully completed in a matter of hours for the five languages of this project. Besides that, the translation company could charge for the software, and therefore obtain a higher margin for this project. Another result of this project is that the telecom company is now looking to expand automation in other fields.

Creating your API

One of the basics of an API strategy is that people, and especially technical developers, are able not only to use what is easily available from technology providers, but can also design and develop the software needed to change the processes. Most translation companies are small to medium enterprises, so this could include outsourcing the development or investing in people to integrate the knowledge and assets. The choice depends on many factors, one of them being the verticals your company is targeting.

After having the right human resources, time should be invested in the process and the API reach. You should open your services to the world, but decide what parts are available and how. In this sense, having a lean mindset will help define what users need. Keep it simple! Innovation also involves rethinking the core of your company — could some translations you provide through the API be free? My fast answer is yes, it could be possible to offer everything you already have in the customer’s translation memory, but there could be other scenarios available as well.

Follow a customer development perspective — this approach involves building a first prototype as soon as possible, even if it is not perfect, and selling it to potential customers, maybe new ones to minimize the risks. Their feedback will improve the product until it is ready for production. This process might take several months, but your API strategy should definitely be profitable after just a few weeks!