How translation can drive innovation in the help center

by | May 28, 2020

Easy LMS client case by Anouk & Co


Software company Easy LMS knew it would benefit their help center and customer experience incredibly if their support articles were published in multiple languages. Still, they were hesitant to structurally translate everything. No more work could be added to the current staff. Budgets were modest. And how to deal with the constantly changing software application?

My company Anouk & Co performed a Translation Quick Scan for them and that made them change their approach to translation radically. Now they create their content in a scalable way, so that publishing their content in multiple languages becomes a piece of cake. Not only did this save Easy LMS thousands of euros, it also prevented them from going into a complex, unscalable and labour-intensive process.

Easy LMS

Easy LMS is an international company specialized in the creation of online education technology. 55% of the turnaround of the company comes from non-English speaking countries. With their product you can create effective and fun learning experiences for your employees or students.

Easy LMS is a company that has adopted continuous improvement by applying the Improvement Kata from Toyota for approximately a year now. In this time they have moved to single item deploy and are now releasing numerous features a day. They run PDCA cycles every day and only solve root causes, instead of symptoms. This brought them to be one of the best LMS software applications in the world, according to, among others, Softwareworld.

The case

In these times where COVID-19 is causing a huge demand for online trainings, the demand for Easy LMS has also grown. This growth in sales also has its impact on the support department.

Easy LMS aims to have most of the support handled through self service. The lack of the support articles in multiple languages was therefore seen as one of the main bottlenecks. However, there was still great hesitation to translate everything.

Jeroen Guldemond, co-founder of Easy LMS, decided to call in an expert, and asked Anouk & Co to run a quick scan, as to get more clarity on the decision making. Easy LMS’ question to Anouk & Co was the following:

What is the ROI of translating the support articles, and how can we make it as high as possible.


“In order to help our non-English clients, our support consultants already suspected that a translated version of our support articles would be of great help, so they started sending out out these articles using Google Translate. They did this manually, so this was very time consuming. Still, making the decision of structurally translating everything was not easy, because we assumed we needed human translators and a project manager to run these translations and we would have to hire someone new to do this.

When I asked Anouk & Co to help me make a decision, I expected them to just make the calculations what it would cost so I could go to my fellow decision makers and have a clear picture to present. It worked out quite differently…”

Results quick scan

Translating the support articles needed to result in an increase of read articles and a decrease of tickets, specifically from clients in non-English speaking countries. The quick scan revealed that two things were considered most important for this new process: speed and scalability.


Given the fast turnaround time of the deploys and the impact that has on the support content, the speed of translation and the speed of change must be very high, as it would severely affect the effectivity if content was not updated.


There were severe restrictions on the impact this process was allowed to have on the organization. It was very important that this new process be implemented with no waste. It was by no means a given that a complex translation process that required additional people would result in a positive ROI on help articles.

The quick scan also showed that for some respondents, there was a preference to using human translators over Machine Translation. This was mostly related to the fact that there were references in the articles to the software application, but also there was a general belief that Machine Translation would not be sufficient for the purpose of the text.


For Easy LMS, continuous improvement is the standard, and they have become very mature in removing waste. Scalability, speed and safety for people are their top priorities. This is how they can offer their clients the best product for the best price.

Therefore, the approach that Anouk & Co advised, was to first experiment what happens if you do not add any people to this process. You do the simplest thing that could possibly work, detect the bottlenecks if it does not, and solve these bottlenecks by assessing the root cause.

This meant that in this case, we advised to work from the assumption that Machine Translation only could be used for translating these articles, until proven differently. This means first adding a number of help articles in a number of languages by hand, and collect feedback from the users. Once bottlenecks appear, look at what the root cause of these issues are, and see if you can solve it in an effective way without adding humans.

In presenting the approach, we empirically showed how paid Machine Translation is good enough for the first step. Also, Jeroen emphasized the necessity to try this out, as this would be the only way to quickly scale up support in all languages. As everyone in the company understands the concept of experimenting and continuously improving, all stakeholders agreed on the approach.

The essential effects of this approach were that:

  • the proposed infrastructure for the first version would be very simple, as there would only have to be an API between the CMS and the MT-engine;
  • no Project Management was required;
  • the (multilingual) support consultants would still be able to own the translations, and this would not be transferred to a separate Translation Project Manager.


We also identified a number of potential bottlenecks, that in our experience often are the cause of decreased effectivity and we proposed a number of solutions, that would take away the root cause of the bottlenecks. After completing the initial test with actual readers, and if it appears that improvements are necessary to be more effective, the following solutions can be considered to improve the process further.

Quality of the source

When testing Machine Translation on the actual content, it appeared that a number of sentences were incomprehensible. When we examined these, it appeared the English source sentences were not well-formed, and not so clear to the English reader either. When these were changed this had an immediate impact on all translations. This is the first step towards better content in all languages. A more structural solution would be to apply writing rules to the source content, which can be supported by multiple tools.


Second, the content contained screenshots. As the application is translated into 10 languages, a support article would only be helpful if the screenshot was shown in the correct language. We advised to look into automatic screenshotting and dynamic linking, so that screenshots would be updated for all languages automatically, including English. This had a positive effect on the quality and speed of publishing the English support articles.

Software references

Third, the references in the articles to the software are a bottleneck, which is not a trivial problem, whether you work with translators or machine translations. Short term solutions include things like circumventing using these references and letting the screenshots speak for themselves, or creating terminology lists and adding those to the MT engine.

However, if you look into the root cause of this, you find that the real bottleneck is the fact that the same content is stored in different places. If we were to look at it from a single source perspective, these references should always be connected to the place they originate, so that any changes, in whatever language, including the source language, will be updated as soon as they are changed in the software application. This is the only robust way to manage changes in the software application, as going through these articles by hand upon every change is simply unmanageable. Our advice was to do more research into applying this single source approach and we are currently close to solving this.

As you can see, these solutions have nothing to do with the translation process, and everything with the way the source content is created.

The results

When Jeroen started looking at the process from this perspective, the real cause of the problems surfaced. He even went one step further, adding content creation to the process of software development. This led to a much faster publication of the source content. First, there was a backlog of at least two weeks before new functionalities would be supported in the help articles, this is now available for their customers when it is deployed, including all screenshots.

In the next step, the process can be automated further. Once this is complete there will be an instant publication in all languages upon deployment of new features.

Results on the level of self service of Easy LMS customers, both in English and other languages, are expected to come in at the end of the month. This is of course the most important question: does translation of articles lower the number of tickets and increase the user experience. And if it does not, what is the cause?

However, even in this first phase waste has already been reduced. Creating the source content has become highly scalable and the time to market is fast. Also, repeated manual labor has been removed in the support process by adding machine translated articles to the database.

But most importantly, without adding any complexity and extra costs, Easy LMS can start measuring the effect of multilingual publications and become aware of the ROI of translation.

What would have happened without the quick scan?

Jeroen :

“We were about to send the content to human translators. However, we would have run into the exact same problems as we do with Machine Translation, but then expecting translators to solve them. Maybe even thinking our translators were not good enough and that would be the cause of the lack of effectiveness of our content. Looking at the process this way really made it clear where to start and how to solve the issues. Not only did this save us thousands of euros, it also prevented us from going into an extremely complex process where problems would be solved in all the wrong places.”

On machine vs. human translation

It could very well be that eventually human translators are necessary in this process. It is by no means my intention to discard the value of human translators. However, taking Machine Translation as a starting point makes you really look at your process as a whole. You cannot ask anyone to fix in a translation what is actually caused by the way you create your source content.

The bottlenecks that were detected in this process were by no means solvable in a controllable way by translators, not even the references to the software application. However, in most cases in the industry, translators are repairing issues that are really caused by a broken process, leading to unnecessary time and costs.


The way translation processes are usually viewed, often leads to unnecessary complexity in processes and automation. Taking a step back and looking at the whole picture can save you a lot of time and money and will lead to happier people and easier expansion to international markets.

The key take aways of this case are:

  1. You can get away with a content creation process that contains a fair amount of waste — until you start translating. Problems in the translation process show you where the bottlenecks are in your content creation process, as it will magnify the problems by the number of languages you are translating into. If you optimize the way you create your content in the source language, translation will be easy. Otherwise, translation can become very hard.
  2. Start simple, experiment and measure. Investigate where your process is not sufficient, and then add the necessary additional elements. If you set up a complex process without a real necessity, it will be much harder to optimize.
  3. If you do run into issues, investigate the cause first, before running to the solution. Only too often a lot more waste is introduced when trying to solve a problem in the wrong place.