Old School Agency vs. Translation Platform

Let’s start by defining what is a translation agency as the translation platform is the digital evolution of the agency.

A translation agency is an organization that takes in files in one language and delivers them in one or multiple translated languages.

Translation agencies are typically powered by tools and people. Most translation agencies have project managers and translators at their core, with project managers quoting, liaising with clients, and placing tasks with translators.

As agencies grow they become more staffed in all of the supporting areas including but not limited to account management, vendor management, quality management, solution architecture, tech, compliance, legal, and any other team that can add value to the overall mission to deliver high-quality translations.

As far as tools go, most translation agencies work with third-party tools. Their core tooling is comprised of a business management system such as Plunet or XTRF, and a translation management system such as Memsource, XTM, MemoQ, or Trados.

More tech-developed agencies will typically have their connectors as well that use available API endpoints to optimize system integrations and workflow. The business management software keeps track of customers, quotes, vendors, job placements, translator payables, invoicing, and other translation processes relevant components.

The TMS (Translation Management Systems) provide the CAT (Computer Assisted Translation) environment for their translators and project managers. The TMS will typically store the client-specific translation memories and term bases and provide translators with key productivity tools such as spell-check and QA verifiers.

So in a typical workflow, a translation agency will:

1. Receive files from a client (either through their portal, file transfer systems or email).
2. Have a project manager download, these files and upload them into their TMS
3. Run analysis and import the analysis into the BMS (business management system)
4. Import the analysis into the BMS
5. Manually add the line items that need to be included in the quote (such as Project Management, or Desktop Publishing)
6. Check the quote for integrity
7. Download the quote
8. Send the client the quote
9. Wait from client approval
10. Reach out to translators to check for availability
11. Place the job with the translators
12. Manage query management by filtering translator questions and relaying them to the client
13. Wait for the translation to be complete and place job with reviewers
14. Fill out amount payables to translators and reviewers provide them with a PO
15. Send files for further processes such as desktop publishing, QA
16. Wait for deliverables
17. Check final assets
18. Prepare delivery
19. Deliver using file-transfer systems or email
20. Manually confirm safe receipt
21. Invoice
22. Handle client feedback

This is SOP. It can become more granular and complex with greater detail and exception management. But the main point of concern for us regarding this process is that it’s:

Error-prone

With so many manual steps it’s only natural to forget to add a line item, to add one incorrectly, to forget a destination language, drop a file, etc…every single manual entry is an opportunity for error and we count dozens of manual entry contact points even in the simplest of projects. You can further enhance the process with more checks and controls to mitigate error but your process will become painfully slow to minimize error.

Slow

Time is lost with every manual step. So many steps require someone’s response or active engagement to move on to the next step. There is little to no real automation potential and the time gaps between each stage add up.

Requires a large overhead  

With so much manual work, Project Management has limited ability to deal with large workloads as they will become stressed, overworked, and make more mistakes when the workload peaks. This setup requires lots of people in different positions to ensure the necessary team elasticity and ability to handle the demand that scales up and down overnight.

Difficult to improve on

It’s very challenging to improve when you have a process that is based on manual steps. No matter how much you train people, add layers, and cross-check you still cannot deterministically address change at a root level. The root level is people and people are erratic by nature (not that we don’t believe in human goodness, but we are not the most consistent of species)

In our platform approach, we tried to simplify, automate and create a technological framework that ultimately does the same as an old-school translation agency: documents in, translated documents out. But we defend that our approach is a lot more predictable, cost-effective, and fast. How in the world do we claim to be better, smarter, and cheaper? Do read on… :)

This is the overview of our process

So right off the bat, we took a process that would take at least 22 different steps to complete a translation process and took it down to 6. That’s a 72% reduction in necessary steps. Each step that has been subtracted represents less room for error, greater process efficiency, and more opportunities to continuously improve the process.

So right now, we have shown that the process is ostensibly the same, but with fewer steps. But as you look a bit deeper, we will examine how the process is wildly different from a paradigmatic perspective.

How a Translation Platform works

Scalable

Our translation platform decides based on parameters and algorithms. It doesn’t depend on people following instructions, receiving the right training, or getting the correct level of handholding. The platform just knows. And it’s not that it randomly knows because a software engineer decided it to be that way. It knows because we have been able to consolidate 17 years of successful agency experience and transform it into software through layers of analysis, architecture, and data science.

Predictable

Our process is predictable. It’s not perfect. But it’s predictable. If we change a quoting parameter we can be 99.99% sure, that parameter will be followed.

Quoting is based on system detected values. Analysis against relevant translation memories, segmentation, and parting are all configured globally for any given client so that no two different projects require separate preparation.

Based on the file type and net word count (word count discounting repetitions, non-translatable sentences, and stored translation memory matches), Bureau Works can provide a quote instantly for over 22 file types. We are still unable to deliver this level of automation for complex multimedia files such as videos and e-Learning but our quoting automation covers more than 90% of the process.

Project Management

This is the beauty of exception-based project management. Rather than spending precious time and attention on things that don’t require time and attention, our platform enables our engineers to focus their time on the things that do require their attention such as complex eLearning projects, client onboarding, and initial customer profile configuration, data analysis and other activities that generate a far more relevant ROI both for us as the service provider and for you, our client.

There is no time lost between file submission and quote. This may not seem like much but we observe across the industry sometimes as much as 2 to 3 business days are lost in quoting back and forth between project management and stakeholders. In today’s world, that kind of time inefficiency has a progressively prohibitive cost. It’s not just that time is money. Time is the difference between making it or breaking along the way.

Job Placement

As you go further down the process the time and management inefficiencies of an old-school agency will only grow.

When it comes time to place a job with linguists with time zone challenges and a non-quantitative approach it becomes a hit-or-miss process.

Often the go-to linguist for that project management group is unavailable. They either plead with the translator to take on a greater workload (which inevitably impacts quality) or they have to go with a non-go-to translator which represents greater project risk. In a world where decision-making is based on heuristics, generalizations, reputation, and prejudice such as “they are amazing”, or “they have such a bad attitude”, making good consistent decisions is nearly impossible.

Project Managers are either working in a tight symbiosis with a few trusted linguists or are out in nowhere land working with people with a very little track record to ensure credibility. And even the credible translators are often working off fumes of past glory rather than recent credit and achievements.

Key Indicators

We overcame this by developing a uniquely quantitative approach to translation performance and quality based on our LEI ( Localization Efficiency Index) that measures how much work post-translation a sentence requires to become market-ready, Levenstein edit distances from machine translation feeds, time spent on each sentence, customer satisfaction polls and other relevant data points that allow our platform, not to just think about who is good, but predicts most likely at any given point in time to perform well at any specific translation assignment based on parameters such as subject matter, tone, and translator track-record.

But that’s not the end of it. Perhaps the most detrimental aspect of the old-school approach is that the high overhead necessary to keep the lights on requires someone to pay for it. That someone is the translator who typically gets squeezed into working at lower rates in exchange for more volume and predictable income. This exchange results in translators having to work faster to reach their desired level of income which means that they are less likely to devote the time and attention that’s necessary to produce excellence. It’s hard to go the extra mile when your rent is on the line.

Our platform approach that minimizes overhead allows translators to set their rates and they become eligible to take jobs provided that they are deemed technically competent and are within reason of basic supply-demand market parameters. This may seem like a detail but it’s the difference between night and day. With as much tech as we bring to the table, translators are still the stars of the show. It’s their talent, creativity, improvisation, and thought leadership that makes the difference between translations that are blah and translations that truly represent your brand.

Conclusion

Traditional approaches lead to traditional results. In our opinion, we believe that it is up to us to not accept traditional approaches driven by the fundamental belief that we can do better, that we must do better. We didn’t develop a translation platform because we thought it sounded like a cool thing. We developed our platform in the attempt of answering this key question: what is the most effective way to predictably deliver cost-effective translations for our clients. While there is nothing wrong with sticking to known patterns and relationships, there is nothing wrong with pushing the envelope and seeing where that can take you.

A well-oiled old-school translation agency can out-perform a poorly managed translation platform any day. There are no perspective one-size-fits in our opinion. But when it comes down to working with content management systems, connectors, continuous localization, and streamlining productivity nothing can beat the potential of a well-oiled platform if your goal is to maximize multilingual content production, productivity, and global customer engagement.

‍