How to Create a Localization Department in a Gaming Company
By: Denis Khamin - Allcorrect
24 March 2020
If localization is a pain in the neck, you're not sure who you can ask to "just do the **** translation," deadlines are always missed, and quality is a lost cause, this article is for you.
If you've read books on HR, chances are they've recommended that you call former employers to ask about your candidate's strengths and weaknesses (especially if they didn't leave on the best of terms). Some authors even take the extra step of telling you to contact contractors the candidate's worked with to see what they think. And when the position involves managing a budget, that's especially important.
But as a rule, it's only recruiting agencies that call former employers, while nobody goes so far as to check in with contractors. And that's why I decided to write this article. It's even more critical given how nobody talks about processes and management in this context.
Why me? Over the 14 years I've spent working in a translation agency, twelve of which focused primarily on localizing games and software, I've gotten to speak in depth with both the founders and directors of gaming and IT companies as well as the managers in charge of their localization departments. Some of our former employees work at those companies, including as localization directors. And that's given me some insight into internal localization processes, while working with external contractors is a relationship we've been part of every day since 2006.
While I organized my experience as case studies based on actual examples I've come across, I did soften or exaggerate spots here and there to make a point.
Localization departments include a localization director, several managers, and possibly the translators who will handle the game translation and testing process.
Really, the answer is that you almost never need your own department. Getting top management involved in setting up the department, poor scalability, a lack of qualified staffing options, potential unprofitability due to erratic workloads, and other factors often lead to a situation where sweeping everything under the rug of general and administrative costs is the only way to avoid facing the music. Well, there's also how excited HR will get when they start collecting the bonuses for all the new hires.
You need a localization department if you're a publisher with at least five or ten major games or unique IPs that are constantly being updated. For all other cases, it's perfectly fine to just have one localization manager or someone else responsible for localization and working with external contractors. But localization management is still something you'll have to take care of if you're developing or publishing games.
Okay, you know you have to do something about localization, so you're getting to work looking for a localization manager or department head.
A major IT company is looking for a localization department head. The main requirements are knowing the basics of web layout and several programming languages, in-depth understanding of the company's products, a solid grasp of one or more foreign languages, and a degree in linguistics. Down toward the bottom of the job posting, there's something about management credentials and a mention of how similar work experience would be a plus.
The only people who respond to the ad couldn't cut it as programmers, and one of them is eventually selected. When it comes to understanding the job, the ins and outs of different languages, and how to work with contractors, not to mention a passion for the job and languages in general, there's nothing to go on. They wanted to work as a programmer, after all. And the quality of the work they do will reflect that.
Here's what a localization manager or department head needs in order of importance:
- General familiarity with the company and its products.
- Experience working with people (translators, contractors).
- Critical thinking and high openness to experience.
- Knowledge of professional software.
- The general concept of localization—awareness of the industry events, participation in conferences and seminars.
- A solid grasp of one or more foreign languages, one of which should be English.
The first point shows that the candidate wants to work for you, the second and third are minimum requirements, and the fifth point tells you they're looking to develop professionally, they like the job. While it takes an interview and a few days on the job to see if the candidate really is someone you want to work with, that passion for languages is something that comes across right away. And there's no way they can talk with all the linguists they'll need to be in touch with if they don't speak English. If they're smart, they'll be able to dig into your product at the level they'll need for localization by the time the probation period is over.
In our experience, it doesn't so much matter what the candidate's educational background is. They could have a degree in linguistics, economics, or even something technical to go along with additional training as a translator.
A good 90% of candidates won't even have three of the six points on the list, though if someone does meet all of the requirements, they definitely have experience in localization. Chances are, they're a professional.
If you don't want to figure out localization and build the processes yourself, look specifically for a localization department head even if you don't have a localization department or your own translators. Need someone with experience? They won't respond to a small company advertising for a localization manager.
From a financial point of view, hiring someone to work remotely or part-time can seem attractive, though that comes with risks—there are precious few people out there who can do quality work outside the office. After all, the job entails getting in deep and interacting with the rest of the team. Almost every company has its own processes localization needs to be integrated into. Really, when the rubber meets the road, the only way for a remote employee to do a good job is if the processes are in place and documented, and there's already an existing team.
Generally speaking, localization managers have their incentives tied to deadlines, localization budget savings, and however the company measures quality.
The variable portion of the salaries earned by game publisher executives depends on the profitability of the games they're in charge of, not to mention their ability to get the job done on time without creating problems for top management. And the localization department is a separate business unit unrelated directly to marketing or game development.
The head of the localization department doesn't take part in buying traffic or analyzing profitability or ad channel performance. Just about the only way they can impact the game's bottom line is by cutting direct localization expenses. And because of that, the company is constantly looking for new suppliers who charge lower rates and selecting a different contractor for each new project.
Each contractor gets fewer total projects from the company, the supplier selection process veers toward complexity and unpredictability, and the company loses its relative importance as an employer. And to bolster sagging quality, the localization department head hires new staff so they can find new localizers, cut out the middle man to go for translators and game testers directly, and keep up with the complex process that is managing that many contractors.
Cheaper contractors tend to require more attention and additional resources from the department managers. Even experienced suppliers with established processes in place try to make up for squeezed margins by opting for less-experienced teams and getting fewer managers involved. And the vicious cycle is complete :-)
The number of localization managers you're hiring is outpacing the number of active games you have, while your translation and testing quality is dropping or just holding steady.
The localization department is gobbling up the resources afforded by upper management and support teams (HR, development, and support for internal services). In fact, sometimes it feels like you're more a service company than a gaming one.
It might be a good idea to spend some quality time with the executives or owners of key partners. You're looking for long-term cooperation, right?
- Take some time to think about whether you actually need one (see "When is it time for your own localization department?"). As a rule, growing numbers of employees in any department points to Parkinson's law, while the quality of work they're getting done is all about the Peter Principle.
- Merge localization and marketing. Companies with mature processes in place bring localization under the marketing roof (just think about the kind of reaction there is when the title of a movie is poorly translated). Not only that, but our research and Google data both conclude that localized games earn an average of 1.9 times more than their competitors.
- Tie KPIs to quality and deadlines. The latter are limited by the physical capacity of the translation team, though it still isn't that hard to make sure everyone's meeting them. As far as quality goes, one good marker is measuring how many times the localization testers send translations back for corrections.
- Correctly calculate the localization budget at least to know what kind of a bonus to pay the department head. In addition to the direct expenses contractors bring with them, the budget needs to include the salaries and bonuses of everyone in the localization department, taxes, paid time off, and the time and other resources spent by support staff, including on facilities, furniture, and technical equipment. All that should be multiplied by your employee turnover rate.
A Chinese online game publisher is looking to localize a game. The deadlines are tight, the company has three contractors, the source content was written in English by Chinese speakers, there's no time for testing, and there are no reference texts in Chinese. Prioritizing the deadline, the translations are handed to three different suppliers, with employees on staff tapped to handle the editing. Only one company suggests using a single software tool to ensure consistency and naming one of the three contractors responsible for complying with existing terminology and managing translation memories. For a variety of reasons, including the lack of time, the decision is made to just go with the existing options.
All three companies get the job done on time, though the quality level is poor. And since editing and compiling the texts needs to happen yesterday, there's no time to figure out which companies performed better or worse. Only one of them gets any feedback. Which, you ask? The one that asked for it, of course. And from what that contractor can tell, the problems that were pointed out are all in sections of the translation handled by the other two.
Besides setting goals, it's important to keep a tight grip on translation quality. The best way to do that, in addition to randomization, is to institute a triple-blind system: the people being checked don't know which part of their work will be checked, the testers don't know whose work they're checking, and the specialist processing the results doesn't know who was checking who. Localization or linguistic testing is the simplest way to find and eliminate errors in the real world.
There are two rules for making sure evaluations are objective:
- The translation and testing teams are equidistant from the developer—if the translation is done by an external contractor, the testing should be as well.
- The translation and testing teams should be separate units with their own independent processes and staff. Translators shouldn't moonlight as testers, and vice versa. That goes for both salaried and external teams.
Linguistic testing is a critical element of the localization process. Even experienced game translators sometimes don't get the context or haven't read the supporting documentation. Alternatively, a Chinese font might have been used for kanji in the game engine instead of a Japanese font, or tags might have been just stuck in due to a software quirk. Whatever the case, the outcome can be problematic. And that's easier to catch when you're playing through the game, making it important to have testers in place rather than dumping the issues on your players.
When it comes to a well-oiled localization process, the manager working for the developer needs a contractor to handle game translation and testing. That's generally two separate companies, though there are some cases where a single supplier with mutually independent teams can take on both sides.
If you have multiple contractors localizing a single project, it's going to be almost impossible to maintain a uniform style throughout. There are going to be different kinds and quantities of issues in the resulting texts as well.
The same is true of cases where you're working with a single contractor, only they have more than three translators on the same project. That forces you to hire two or more editors who will have different linguistic styles and mistake profiles.
Just like in Brooks' The Mythical Man-Month, linear expansion in the number of people working on a problem doesn't exactly net you a similarly linear reduction in the project time frame. It can even start pushing deadlines back once you hit a certain number.
But what do you do when deadlines are tight? That's when it comes in handy to have a single glossary and a general translation memory that have to be steadily updated. And while either the developer's localization manager or one of the contractors can take responsibility for doing that, it does need to be just one person. That will cut down on errors, though it won't eliminate them.
And so the perfect translation is handled by one translator and one editor working within reasonable deadlines. That's about 10,000 words a week.
Company Y uses GitHub for file storage, its translation resources are stored in a separate repository. Crowdin automates the localization process by uploading modified strings to the project for translation. Once the translation is done and confirmed by the localization manager, the text is uploaded to the same repository with a language tag. After that, it's automatically incorporated into the build that's sent off for testing. The process leverages the API for multiple solutions, the company's own platform, and a single manager.
That workflow delivers the translation and testing of updates at record speed regardless of their size, and it's suited to work both with freelancers and organizations. It saves the company money, too.
On the other hand, maintaining operability means regularly updating the software on your own and testing out any new functionality to make sure it fits into the process. Managing and upgrading the flow is a job for a professional localization manager, someone who knows all the nuances of every piece of software involved. And that weakness in the chain shows up when they quit, take maternity leave, or have to move on for any other reason. Finding someone to take their place can be nearly impossible (see "Assembling the team").
Below you'll find a simplified visualization of the localization process pulled from our Culturalization Guide.
Unlike the similar layout in Nikolay Bondarenko's article, ours is much simpler. Today's companies leverage automation to launch translations into all languages at the same time (the same goes for testing).
We can break the automated localization process down into four segments:
- Automating text import and export. For example, there are Unity extensions that import texts directly to Google Docs.
- Automating the localization process. There's special translation software available for that, both server- and cloud-based. We use memoQ, Trados, Smartcat, and Crowdin depending on the client's needs and the kind of job we're being asked to do.
- Automating the link between text uploading/downloading and setting localization objectives. This tends to be the most challenging part, as game development is anything but standardized, and the range of solutions different companies use for the various dev stages can be expansive. Still, their API and ready-made cloud solutions can be leveraged to automate the localization process fairly quickly.
- Automating testing. It depends on the platform and other particulars, but localization testers generally go with one of the bug tracking systems out there. Jira, to take one example, is our go-to option.
But the simplest and best way is to have your localization contractor integrate you into the processes they already have in place. Serious companies will offer multiple ways to make that happen, giving you the strengths and weaknesses of each.
And of course, if you don't have a task management system on your end, and you're just sending game texts back and forth using a network drive, don't expect to find a silver bullet that will solve all your problems.
We've worked with American, European, Asian, and of course Russian clients. And while in our experience a businesslike approach and passion for gaming always outweigh cultural differences, there are still differences to be found between those different cultures. For example, the only way you can build a trusting relationship between client and contractor in Asia and Russia is if the directors or owners both know each other personally and are involved in the process.
Company Z develops and publishes mobile games in 14 languages. The new localization manager has their own pool of contacts consisting of several translators and localization companies for each language pair. And because they're looking to trim down the budget, the manager hands each language pair to a different translator, selecting the cheapest options before checking and compiling the result in a file on their own. The file is then sent off to the developers. Testing is handled by a separate company, though tight deadlines mean there isn't always time to get them involved.
Company Z saves up to 30% on localization (compared to the cost of having the entire job handled by a single contractor).
The localization manager never has enough time for the mountain of work that comes with handing out new jobs, evaluating the results, and lumping the files coming in from all the different suppliers into a single text. Because of that, the wrong languages are occasionally uploaded to the game, and there are often other mistakes that slip through testing. The manager can never take a vacation—they're the only person who can manage the pool of suppliers. And if they quit, the whole process has to be rebuilt from scratch.
The app description and interface were translated into English by an employee who speaks the language. Once that was done, the job was handed to a localization company, though no final check or linguistic testing was done for the completed texts.
The app was rejected by Google Play moderators due to English mistakes. Once shareholders had dug into why the app was blocked, they laid the blame squarely at the feet of the contractor.
Case 7. How to work with contractors
A major Asian game developer and publisher handled localization with in-house translators. After a little while, the job got to be too much for them to handle, and the department was reorganized. Each translator was responsible for localization into their language. An external contractor was also hired for a big project that entailed translating a game into twelve languages simultaneously. As changes were made to the game text throughout the dev process, the in-house translators found out about them at different times. Each sent orders to contractors independently. Of course, everything needed to be done ASAP.
For the contractor, that looked like twelve jobs coming from twelve different clients with slightly differing requirements. And when they came back with questions for each of the twelve languages, the in-house translators would even give answers that contradicted each other.
But it was still just a single project, with a single correct answer for each question.
The time everything took was multiplied by a factor of twelve on both sides, for the game publisher as well as the contractors, while deadlines were pushed back, quality dropped, and expenses skyrocketed. And the publisher might not have even noticed how high those expenses were climbing—see "When is it time for your own localization department?"
On the plus side, the in-house translators could use their own time to translate small texts and do spot checks to see how the external contractor was coping with the job.
- Don't let the number of contractors you're working with get out of hand. Working with you needs to be profitable for them.
- Run all localization and testing jobs through a single point of contact on both the technical and human sides. That's one source of texts, one work environment. One person making decisions on your side, one person on the contractor's side. That will streamline the process and drive quality.
- Get to know your contractors. We always offer company executives the chance to meet with our upper management or founders. Having a strong relationship lets you find out how your project is coming without second-hand embellishment, also giving you another lever you can pull as needed.
- Give feedback. All localization companies make mistakes—the difference is in how many they make and how quickly they learn from them.
- Read through what suppliers send in response to your feedback. While the contractor is risking a significant amount of money, not to mention their reputation, your employee is at most putting their bonus on the line.
In line with a lower cost of living, companies in Eastern Europe spend far less on salaries than those in Western Europe or the US, which means you can get more bang for your buck with them. But you always need to see how they react to feedback. If they ignore it or throw it back in your face, it's time to think about a replacement.
Feedback should be a regular feature of any project. And at the end, do a review. Without something like that, even the best localizer won't be able to help you. Serious companies offer detailed feedback for all projects every month, thoroughly studying the responses they receive.
Over six years spent working with a global AAA video game publisher, we built up an enormous pile of feedback that taught us how to improve our work. A lot of evaluations, on the other hand, we questioned and ultimately had reversed. For example, we get monthly feedback for all 17 projects and 12 languages we handle for them, and that gives us a never-ending stream of ideas for how we can get better.
EA's localization department is manned by just 10-15 employees.
As it turns out, I ended up with an article on how gaming companies can build their localization process. Just like any other process, localization is something you have to set up, manage, and constantly improve.
Perhaps one of the most surprising takeaways from this article is that a good contractor is more valuable and reliable than a good localization manager. Once you've built a strong relationship founded on trust, your localization partner won't burn out or run off in search of greener pastures. Instead, they'll build, maintain, and improve how they work with you; they won't get sick or take time off; and they'll be better qualified than anyone you can hire (with the exception of Tony Localistark, of course).
Here's a quick test to check for
coronavirus healthy localization processes, or how to tell if everything's working well with your localization.
There are just two things to look for:
- Replacing the localization manager, department head, or one or more contractors doesn't cripple the entire process. You can see how this works the day your localization manager starts their vacation.
- Your localization quality is getting better or at least holding steady, and the number of employees you have working in your localization department is either remaining the same or growing slower than the number of projects they have to work on.
Find some good managers and contractors, and there's a chance everything will work out for you. Maybe.