Translation Workflow Automation: Everything You Didn't Know You Wanted to Know

A new installment of #AskTheExperts, a GALA series where we ask translation industry experts inside and outside the GALA community for their insights and advice on managing business processes and digital transformation. 

Today GALA expert, István Lengyel (CEO of BeLazy) shares his insights on translation workflow automation.

 

 

For more from GALA Experts, subscribe to our YouTube Channel

 

Whether you're new automation, or whether you're already an advanced user of automation tools, watch the video to the end. István Lengyel has some very useful insights to share.

And here is a recap of what we discussed.

Building BeLazy 

"BeLazy is a new venture and it's also a new type of company. I've been working very hard to understand those areas that haven't really been addressed much. I remember that at the GALA conference in Munich, I had a conversation with Patrick Nunes about DTP and about the fact that nobody does presentations about DTP."

"I also learned that there are many areas about project management, outsourcing and processes, that people don't really have a good understanding of. At BeLazy we are addressing big topics like scalability, availability, SLA compliance. We are trying to create a company that's actually connecting the links that are missing from the industry for the supply chain automation, rather than reinventing the wheel. I am working with many, many fantastic companies (especially LSPs) that want to get rid of the busy work and move one level up in maturity. But what is interesting is that as they want to move one level up, we are building the stairs for that level of maturity that they want to reach."

What does localization maturity mean?

"Localization maturity was something that that Common Sense Advisory introduced a good while ago, and it's still pretty much valid. There is also a maturity for translation service providers, which is very similar, but obviously focused on different processes. The maturity itself, now, refers more to a process than to a company, because a company has a mindset of wanting to be mature."

"I want to make a distinction between a mature and a big company, because I've seen very mature small players and I've seen very mature big players as well. What is really important today is to define the processes and the organization, and also to work with objectives. If you decide to design your processes with the objectives of each step, then those processes are usually right. A couple of years ago, we've seen a lot of nonlinear work. Today, because the content has been decreasing in size, we have arrived at the stage that it is possible for the majority of the companies to linearize the workflow."

What does automation entail in the translation industry?

"In my experience, there are three kinds of automation in the localization industry. There is the linguistic automation: this is obviously the MT-related automation as well as the MT quality estimation and other simple things, like changing the number format. Then there is the process automation offered by different tools. And then there is a third kind of automation, the transaction automation that happens between systems. This is when connectors and middleware come into play."

"Most of the time automation is very simple: list all the possibilities, and then build the rules. People talk about AI (AI or machine learning, I prefer to call it machine learning) and that requires data. At the moment, for our industry, this is just not worth it, from the process perspective rather than from the linguistic perspective. "

What tools can automate the workflow?

"There is a wide array of systems. I would start with the supply chain because the supply chain starts at the top, which is an enterprise usually or an NGO or whatever it is. And traditionally, the procurement of these companies has pushed translation companies towards a single monolithic big tool, which was the enterprise translation management system. In reality, using one tool was always hard, because translation is provided by a supply chain, not by the company that buys it. It’s hard to keep efficiencies and use many systems."

"With the cloud, with the appearance of new systems where content is today stored and managed, keeping just 1 single system is often not possible. So, first we have seen business management systems, vendor portals and translation management systems as well as some proprietary tools (some developed by LSPs). Now, we see development-related tools (like Lokalise), automation orchestrators (Zapier, Integromat) and content sources like WordPress/WPML, Magento or, like recently, headless CMSes like Contentful."

"Lately, we've seen a growing number of tech companies that produce a lot of commercial translation technology, but the problem is that there is often no connection. I hate it when somebody is asking: "Does this have an integration checkbox?" An integration is never a checkbox. Integration needs to be use-case based. Unfortunately, use cases are not discussed enough."

"There are other tools, of course, coming into the picture that are part of the translation process. I'm talking about financial tools, reporting tools, invoicing, tools, payment tools."

What's the first recommendation for introducing automation?  

"My first recommendation would be to map out what you're using. So, make a list of your customers, make a list of the customer workflows, make a list of the customer workflow steps and then map the systems and the technology together. So which kind of tools are are supported? In which kind of systems? What are the workflows, so which systems do they interact? And, of course, don't forget to define your processes in detail, down to the level of who does what and when."

What can you really automate in the translation workflow?

"The typical answer is: the small repeating jobs. Because small repeating jobs are linear. There is another answer, which is: workflows with relatively few exceptions, where you can enumerate all the exceptions, all the variations. These jobs should be metadata-driven."

"The rule of thumb that I would like to give is the following: try to think about it as a new colleague. What people know what to do by only looking at the data available, and three people would do the same thing after a training course ...that can be automated."

How long does it take to introduce automation?

"It depends very much on how well-defined a problem is. Anywhere between half an hour to a year. Buy-in, fear of change are things that vary from one company to another. The number of people involved, and the length of attention also matter."

The ROI of automation

"It’s simple: the problem should be costing more than the solution together with its preparation and upkeep. The first thing is: remove variation introduced by humans. That’s cheap but needs buy-in which is often hard. Testing and understanding often goes hand in hand. Automation is good to standardize, to allow giving attention to other things, and to speed up. These three objectives in themselves are already worth examining. Automation also has an additional benefit, which is standardization. This allows people to pay attention to new things and to speed work up because you don't have to wait for individual decisions to be made."

What's the impact of automation other than increasing profitability?

● Scalability: Automation allows you to take on more projects by setting up rules to automatically accept and approve certain types of jobs even if nobody is in the office.

● Sales: Sometimes the key to winning new customers is not additional translation services, but a stronger technology proposal and more process optimization, or a certified process.

● Translation is never about one company or one vendor, it’s about a supply chain. Companies that speak eco-system are more likely to get attention.

"Also, when equipped with technology, project managers don’t have to allocate their time and effort to tedious and monotonous tasks, such as copy-pasting and file downloading. By helping them focus on tasks that require more thought or creativity (like client-relationship management), an LSP increases job satisfaction and their confidence when handling increased throughput."

"Finally, automating translation workflows requires clear process thinking, which ultimately translates in reduced mistakes and more accurate deliveries."

GALA Content: What’s In It for You?

Why invest your time in reading GALA’s webinars, articles and blog posts, and checking out the webinars? Because they’re food for thought as you evolve your business strategy

Sign up here for our newsletter on globalization and localization matters.