Convincing subject matter experts to use CAT tools

In terms of specialized translations, subject matter experts (SMEs) are the last line of defense. But if you’ve ever tried hiring professional engineers, financial or marketing managers to review texts, then you know that these people are generally busy, and they prioritize their work according to their own rules. It is challenging to convince them to follow translation company workflows and to use specialized software. However, a UK-headquartered language service provider (LSP) overcame this challenge by building a process with 70 experts working alongside translators in the same ecosystem. Here is how they did it.

The story begins in 2013, when a global automotive/aerospace engineering company changed its translation provider because of quality issues.

The problem came from a fragmented workflow: the engineering company employed in-house SMEs who worked separately from translators hired by the LSP the enterprise was using at the time. Experts referred to Microsoft Office apps, which they knew well, while translators used computer-aided translation (CAT) tools. Because of the different ways of working, project managers were supposed to import changes manually. However, it was often the case that the managers didn’t import the changes. Under pressure to deliver on deadline, the LSP often added corrections to the final product only. They forgot or simply did not have the time to update translation memory (TM) and term bases.

As a result, translators repeated mistakes over and over again. Review changes did not recur in the subsequent texts, and SMEs had to redo their corrections every time. It meant a waste of their working hours. At first, it seemed like a minor problem because the final translations were good. Nevertheless, over time the client had to terminate the contract with the LSP.


Lower tech requirements for the validators

The Translation People, a provider of language services headquartered in the United Kingdom with offices in Europe and the United States, approached the engineering enterprise with an offer to connect SMEs to translators on the same platform. This way, corrections could be added to the TM in real time, and emailing files or manually importing changes could be eliminated. As the TM grew, SMEs would have less and less work, freeing up their time for other activities.

The experts, or validators, comprised a diverse group of more than 70 people scattered across numerous locations and time zones. Some of them were stationed in the United Kingdom, but others were based as far away as South America and Asia. The engineering company conducted translations into 25 target languages with two to four reviewers per language: factory directors, marketing people and financial managers. They edited translations in their spare hours, often as a favor rather than as a full time job. They knew terminology in their field really well, but generally were not familiar with translation tools or processes, and had little time to learn them.

For SMEs to start working in CAT tools instead of Microsoft Word meant that the translation company needed to overcome the SMEs’ resistance to technology. The LSP chose software that had a lightweight interface and operated directly from the browser. The software did not require installation or updates and it was possible to run it on any machine without having to lift corporate restrictions on new software. Furthermore, it allowed the LSP to provide licenses to translators, and use the best people for the job, even if they didn’t own desktop translation memory as was often the case with linguists in developing countries and emerging markets. After these benefits had been brought to the negotiating table, the LSP was awarded the contract and started its work.

The LSP began with standard steps such as testing technical translators, as well as building TM and glossaries from aligned texts. However, the biggest task of training SMEs still lay ahead.

To organize this, a project manager set up a dedicated web page on the LSP website. It contained a link to the agency’s project management system and a second link to the CAT tool for validators. In addition to the webpage, the project management team created a PDF reference guide for SMEs. It contained instructions on using the software tool and a step-by-step description of the workflow.

It made sense to invest time into making a proprietary guide, rather than to refer the client to an official wiki. Coming up with proprietary  training materials increased brand awareness, since The Translation People representatives were new to the employees of the client. They needed to establish themselves and make their name known. A concise guide that contained only necessary operations was a great vessel for this.


Training and communication

When the landing page and the PDF guide were ready, the LSP started training. With the audience scattered around the world, sometimes as far away as Japan and South America, it made sense to run the sessions online in webinar format. Each session began with an introduction of the company and the process, and followed up with a live tool demo on shared screen.

From 10 to 12 experts attended each session, grouped by time zone, languages and availability. It wasn’t easy to bring all SMEs together, and in the end the translation company ran seven webinars to reach everyone.

The sessions went smoothly, but attendees didn’t ask many questions. It was only three months later while reviewing and validating texts that it became apparent how few had absorbed the training. Even though the chosen software system, Memsource, is relatively easy to start working in, the idea of TM, with its bilingual format, segments and suggestions, wasn’t quite clear to SMEs. Project managers received many questions and they could see that some of the validators downloaded source texts to work on them locally. To overcome this difficulty, the LSP had to update training materials and do a lot of individual training.

Faced with some resistance to technology on the part of SMEs, the LSP had to stay on its feet and adjust training. They released video demos with instructions and created a shorter guide, only one page long, to complement the previous eight-page document. The new guide was handed to the SMEs as a printout for their desks.

After the second round of training and the distribution of the one page document, the number of technical support requests came down by about 40%. Obviously, the experts didn’t have the time to scrutinize a longer PDF guide, but were quite happy to use a brief printout.

As a final step the communication approach changed and project managers focused on explaining the basics of translation technology to validators who still experienced difficulties.

They explained the benefits of TM for the enterprise, and specifically for the SMEs themselves. Over time, the use of the CAT tool led to less work on the part of validators because their terminology and style were being taken into account. They were visible to linguists when they translated, which led to a general decline in the number of corrections.

SMEs saw that this work was rewarded with improved translations, and were motivated to follow the suggested process. When proficiency with the CAT tool approached 100%, the LSP disabled file downloading to make sure validators worked only online.

In the end, it was not a single specific action that changed people’s minds, but a combination of factors. While training sessions and material helped, and a lightweight interface made it easier to learn CAT tools, it came down to personal motivation, general practice and familiarity with the tool over time that did the trick.


A 25% efficiency increase

Now, two years into the project, all SMEs are working in the translation environment. SMEs implement corrections directly to a centralized TM, and it continues to grow.

On average, the number of changes per job has decreased by about 25% across all the languages. For a team of over 70, this means dozens and hundreds of saved billable hours. Peak improvement values in some languages, where the quality of previous translations was lower, reached roughly 45%.

Although it was a successful project, in hindsight, it would have been better to run live or face-to-face training sessions, even if it was costly. This would have established presence and the correct approach from the start. Another thing that could have been done differently would be to run the training right before the first project, leaving less time for skills to be forgotten. A pilot project or a dummy run could also have been used to solve this.