Optimizing localization processes with Willian Magalhães from Quandoo
26/07/23
The first episode of the Localization Process pod! In every episode, we'll be learning from one guest about the way they do localization for digital products. Our very first guest is Willian Magallian, Senior Content Designer and UX Writer at Quandoo. We'll be hearing all about Quandoo's unique process and the journey to make localization better, constantly.
Audio
Text-based video
About Willian Magalhães
Willian Magalhães has been working with UX Writing since 2017. Currently, he is a Senior Content Designer at Quandoo. Brazilian, based in Berlin, Germany, Willian is passionate about UX Writing and uses language to execute business strategy. Prior to working as a UX Writer, Willian taught English as a Foreign Language for over 10 years. He holds a bachelor’s in English Linguistics and in Systems and Analysis Development. He’s also been a UX Writing teacher since 2021, having taught in universities and in UX Design Schools in Brazil.
Enjoyed the Localization Process Pod? Subscribe here, on LinkedIn, or on the Localization Station website to get notified about the next ones.
Transcript
Michal: Hi there, I'm so glad to have you here for the very first episode of the Loc Process pod, a podcast dedicated to the localization process for UX and digital products. We're going to be learning from the wisdom and experience of others to optimize our own processes and produce better experiences.
In this podcast we'll talk with product leads, UX writers, product designers and more to learn what they do to create incredible localized experiences. Learning from each other can help us create better industry standards, and better localization processes for digital products worldwide.
Today I will be talking with Willian Magalhães, Senior Content Designer and UX writer at Quandoo. Willian came to Quandoo as a content designer but also helped recreate and redesign their loc processes to streamline, scale, and bring better experiences to audiences in other countries. We'll hear all about it from Willian today. So, Willian, thank you so much for being here!
I would love to know if I'm pronouncing your name correctly. Is it... willian Magalhães?
Willian: Yes.. Willian Magalhães. Yes. Very nice!
Michal: I actually looked it up before. I'm just really used to people mispronouncing names, I guess, so I had to look it up myself first. So, welcome, Willian Magalhães. We are here to basically discuss how you have built your localization process at your company, at Quandoo.
And you've mentioned that you're now localizing into eight languages. Is that correct?
Willian: Yes.
Michal: That is amazing. So I would really like to hear from you first before we kind of start digging into, I guess, the meatier part of the interview, about the processes themselves. I would love to hear from you what kind of background you brought into this role and kind of where you came from into it.
Willian: All right. Thank you so much. So let me tell you a little bit about me.
I have a degree. I started my degree in English linguistics And then I taught English for 10 years in Brazilia. I'm from Brazil. And then, after 10 years in the classroom, I was thinking, it's time for me to do something that I also love, which is technology.
I didn't know what to do exactly. I just like thought: I should get into a... I don't know, a university course or something to give me some guidance so I could find out what it is in the market I can do. So I started this university degree again, the second one. In analysis and systems development.
In my first semester, I was lucky enough to get an internship at IBM. And inside IBM I discovered Watson. I started working with IBM Watson, like building chatbots and working as a conversational designer, UX writer, since then, that was in 2017. And then after my internship, I started being contacted by companies. Hey, could you come and help us with our chatbots and conversational design. So this is where my foundation is, UX writing for chatbots since 2017.
And then, two years ago I joined a global company called Cinch. And I was able to start working with international teams. Because we had people from all over the place, and because I can speak English, and people in my team couldn't, I was the one reaching out to them, like getting information, bringing information to my team. So the interest of having these connections with international teams was born there.
And then, after Cinch, I was invited to come to Quandoo here in Berlin. I'm based in Berlin now. And the main objective, the main goal of coming to Quandoo was to help the company with localization because we had sort of a process, but it wasn't something documented.
It wasn't something that followed a flow. It was something that was like, let's create a ticket, and let's follow this ticket until it's done.
So when I first came here, I tried to understand the resources that we had, the languages that we localized, and which kind of markets we were localizing to. It took some time. For example, in 20 days, it's gonna be a year that I'm here, but it took around seven months for me to really start doing something about localization. I believe the first six months, of course, onboarding and learning the product and understanding users', data, users' behaviors, all of this counted.
So recently I've finished the localization process here in Quandoo. It's a mix of having localization specialists as externals helping us in this flow, and also counting on some local employees from Quandoo to also help us out in this process, because we have people from 52 nationalities in the company.
Michal: Oh, wow.
Willian: Yes. We're People all over the place so we can, you know, rely on these people to also help us. I know they're not localization specialists, but it's better than me, that I don't know the local customs of that place or languages. For me, I cannot tell if that sounds weird or not.
Michal: Yeah. Okay, so let's start with like the beginning, cuz you said you came on to Quandoo, and you did already localize at the company, but you didn't really have a set process. Anything documented . And so I'm curious to know how that went. Before you even started streamlining and improving and optimizing.
Willian: All right. So the original process was like, we had this LSP, this language service provider here in Quandoo we use Crowdin.
Michal: Mm-hmm.
Willian: And we also had Google spreadsheets and we were translating the content in a spreadsheet. And the content that was translated in a spreadsheet was then by a UX writer inputted in Crowdin.
And this was the process, I think people created a spreadsheet and asked, can you translate this?
And then after it translated, you would be able to go to Crowdin, find the keys, and then update the strings with the translations that you got.
Michal: Mm-hmm.
Willian: And then sometimes we were requesting translations from from a provider and sometimes we were asking directly people to translate them.
So there wasn't like a rule, like, for this, we should follow this path. And for the other thing, we should follow the other path. There wasn't clarity in this process. So what I did, I was like, okay, so let's understand first what kind of content comes to us so we can translate. So it's a small, like, content audit.
Is it a paragraph? Is it an article? is it a UI text? Where are we gonna use this content? And then, is it a marketing campaign? Is it a product thing? So this small audit is done now because, depending on how big the text is, it's very expensive for us to keep requesting them from translation providers, you know, so depending on the amount of words we have this flow. We may get in contact with them earlier to be translated because it's a very big thing, like an article or a page or something, a tutorial. Or if it's a UI copy text or something, we ask them, just review what we got from a translation service.
Michal: Right. So basically what you did was you kind of started categorizing the types of content that you have, right? So that you can find a process that is more optimized for each kind of content.
Is that what you did?
Willian: Exactly.
Michal: Yeah.
Willian: And not only that, but we eliminated the spreadsheet step. Because I gave power to the translators to work directly inside our LSP.
Michal: Mm-hmm.
Willian: So I gave them the role, as translators, inside Crowdin, and they're free enough to go there, edit, change, save, update, vote, whatever they feel about the copy, they can do.
It's like I made it scalable. I scaled this process, because I believe that the success of this process is getting me out of it. Like as long as it flows without me, I'm good. And that's the core thing for UX that I believe in here. Like, scaling people to do your work without having to call you because they can't do it.
You may be like the consultant, that signs off. This is good. Nice. Like, I mean, I will, you know, do something, of course. But for example, empowering people to write, which is my role here, UX writer. Empowering people to write, empowering people to start talking about localization in their teams.
And I am having this already. I presented the localization topic here. I actually had two presentations in the company. The first presentation I delivered was a very educational one, like talking about the topic. What is it? What can we do? What are the impact?
Let's see some examples of wrong localization and right localization, because if we have the intention to go global, we should be worried about global content as well.
Michal: Yeah
Willian: So the first meeting was an educational presentation for the company, for POs, PMs, developers, designers, and everybody else.
The second meeting was technical. The second meeting was like, okay, so let me present you Crowdin. Let me present you the scheme of the flow. Let me present to you, like, when you can get in touch with me. Let me present to you what we can do in our teams. So after these meetings, I got people from other teams getting in contact with me saying, hey, let's localize our website.
I mean the homepage for this region. For example, we are in Hong Kong, we are in italy, and they're very different markets, but we use the same image for all of them. So they started like, hey what if we change the picture here to reflect more of that market? And then I was like, yeah, that's localization. That's what we're supposed to do. How can I help you? So. I think I've planted the seed and it's growing slowly, but it is. People are actually reaching out and saying, hey that's really nice. How can I do it here? So let's talk, let's have a meeting. So this is the outcome that I expected.
Michal: So you became like the localization advocate, right, at Quandoo?
Willian: Yeah.
Michal: Were you the only localization advocate, did they have other people handling localization as a managerial role?
Willian: I don't think so. I don't think so. I believe this localization task was spread among employees.
Like for example, when I go to Crowdin and I see the history of translations, I see names from people that were once product managers, and I see people who were content designers and I see people who were product owners, so I believe there wasn't this responsibility of the area given to a person, so this person could, you know, organize and manage who will actually review the translations, who will actually do the translations, where we gonna get the translations from.
So, yeah. I believe it wasn't given to a person to manage this area.
Michal: Do you feel that having a localization owner transformed not just the processes, but also the results that you got?
Willian: I believe so because we are getting outputs on this.
It's too soon to get some data because we have just implemented the process. But we need to talk like designers now. How to measure – and if we can measure – like how can you measure user satisfaction on seeing an image that reflects to their culture?
We know we are in the right track, but I mean, we still have to sit down and say, hey, okay, we're localizing. So what? So what are we gonna do with it? I think this is really important and it's a very strategic conversation we should have. And then after we have this conversation, I think we're gonna have more material to keep on, advocating for the discipline.
Michal: So now I wanna start talking about the process itself. So basically, at what point do localizers come into the picture? Do they get, do you consult with them earlier on before you really start actually creating content in the first language? Or do they only come into the picture at a later point?
Willian: Okay, so the first thing that we request in this localization process is, whatever yOu want to be translated or localized, we need the strings. We don't work with Slack strings anymore. We don't work with Google Sheets anymore. We don't work with this.
We need to have this in a place that we have control. So Crowdin is the place for us. If you want us to translate something for you, what you have to do is to create a ticket on our Jira board. We have a localization board now. And we have some instructions on how to create this ticket. Create a ticket, put the Crowdin links, give us some context, give us some deadline, and we will then move forward with this. So, When they create this ticket and they put the strings, I see what languages they're requesting. And then I see the context and I try to help the localization specialists. Because they're not in the teams.
These people, they're like consultants. But they're not in the teams, so they don't know what we are doing. The product we are working on, the feature we are working on, so they need to be contextualized. After I try to contextualize for them, I request translations from the translation service that we use here in Quandoo. And then this ticket moves to translations from this service going on. As soon as it's done, we go to the second part. Actually, the third part, which is localization specialists will sign off this content, so I let them know that they should go in Crowdin, see the translations we've got and sign them off or change or adapt or see what we can do.
So once they're done with it, then we are able to do quality assurance of this content. So we build whatever we are testing, we go into test environment. We see how it looks. And then we send it to production according if it's right or wrong.
Michal: So basically you do design, it goes to development, and then the people who create the keys are developers, at that stage, at the development stage.
Willian: Exactly. Design designs. I write the copies in the design. After the copy, and that part of the process is done, front end developers build the strings on our LSP. As soon as the strings are created in our LSP, they come back to us for localization.
Michal: So let's say, basically you do the design, you do the writing, you send it to development, it goes to localization, and then the localizer comes back to you and says, oh, I'm sorry, but I need more than one line for the title, or my language is longer so it doesn't fit, or something like that. You need to make any adjustment to the design. So what do you do when that happens?
Willian: Okay. So in this training that I gave, the educational training about localization, I talked about internationalization.
Which is pretty much this, and internationalization is under my responsibility because I am the one creating the copy in the first place.
Michal: Yeah.
Willian: So I'm the one who has to keep this in mind. I should understand that when we translate, for example, from English to German, German is going to be smaller, because in English, we separate the words more. In German we tend to get all the nouns and make them compound nouns.
So they are usually smaller. However, when I translate, for example, from German to Chinese, I know that Chinese is gonna be even smaller. So these moves that can happen inside the ui, I have to have them in my mind. So as I'm the one creating the copy, I'm pretty much preparing for how it could be when we get bigger or shorter strings.
We don't do very long CTAs in our product. Usually our buttons are safe because of the internationalization techniques. But the others are just paragraphs and titles and they can grow because... yeah.
Michal: Yeah, absolutely. Flexible layout can definitely make this easier.
What kind of response did you get when you started creating and outlining everything? You obviously maybe stepped on some toes, I'm guessing people had a way of working and they were used to it, and now you are starting to change things for them. So what kind of response did you get? How did they react?
Willian: They liked.
Michal: Yeah?
Willian: They liked, because in the past they were translating. Localization specialists, they were translating. Now they're signing off what has been already translated, so it's less work for them. So they liked because It's less work, it's faster. So for them it was, I believe it was good.
And for POs as well. It's easier to find a string in Crowdin than find a spreadsheet in a drive that belongs to some person.
So having this content all in one place is also good for organization. And for example, in a case that an employee leaves, that spreadsheet is going to be deleted. With his or her account. But when you're talking about LSP, everything's gonna be there.
Michal: All right, so now you are maintaining your source of truth on Figma, right? And then,
Willian: Yeah, for some designs that we create, yes. And for some other things that we don't have in Figma, Crowdin is a source of truth.
So
Michal: how do you keep things from getting confusing? Like if I update something in Crowden and then something changes on Figma and then they're kind of not in sync now, how do you keep them from being mixed up?
Willian: As we are a very small team, and I'm the only content designer for B2C, changes for strings usually go through me. We like design systems updated. We don't have this, many changes in our buttons or anything. When we have something that's going to affect the whole product we very well aware of what's happening.
So for example, we recently changed our main CTA, from "reserve" to "book". When changes that big, we are usually aware and we update in Figma, Crowdin. So it's okay. And we have the history in Crowdin, so it's okay.
Michal: I'm curious to know, you said you were doing design, you're doing development, and then you're taking everything to a test environment and you have it sent to the localizers again for review.
So how does that go? The quality assurance process?
Willian: Okay. Quality assurance process is pretty much done by QA people here. And we don't check the content again. We only check for bugs or visual bugs. Okay. For example, we translate it in our LSP, we localize in our LSP. QA understands that and sees that, hey, this is not good visually. What's wrong? It's not like in the design. So probably QA will compare this test environment with our Figma file, which is the source of truth. And then if necessary, we can contact the localization or design and then we can use any workarounds to, to fix that.
Michal: So you have all these processes basically set out now. How do you plan to proceed? You have everything outlined, you have your documentation. What's the next step for you? How are you going to kind of further optimize this?
Willian: Yeah, so to be very honest, we need to clean our LSP, because it's pretty much full of sentences we don't use. And sentences that are reused. Because of this the localization specialist get lost. Like, what should I actually translate and verify? What was created in the first place? Or what the translation is saying? Yeah, so this is like the thing right now. So maybe cleaning the LSP right now, so we can have a better environment for localization specialists to work for. But in relation to the content that is being created, I think it's pretty much working well. This process has been working well.
And something that I learned in my career is that, I helped create the design process here in Quandoo, and I've created the localization process. And what helps the most is create a process from scratch. Don't bring this shelf processes to your company. Because you may not have the resources, you may not be able to follow this specific step in that process and then it's gonna be a mess if you try just to apply this shelf process that you get from whatever person created or whatever company created.
So what's been working for us is like, we just sat down and we said, what do we have? How can you do it, you know, in a quick way? Because usually, we have deadlines. Strings are done, they need to be translated as soon as fast because they need to go live. So I don't have much time to keep, like, I brainstorming with our localization specialists, like, what should we do here? I'm so sorry. They have to look and sign off.
Michal: I'm really curious to know though, you just did this, so maybe you don't have data yet, but if you make a change, like going from "reserve" to "book" or from "book" to "reserve", do you always make that change throughout, across all languages? Or do you check that maybe some languages will be better off with " book" or do some languages even don't have "book" or "reserve", they have another version of the copy that is unique to that language. And if so, do you keep that or do you change it as well?
Willian: Then we involve the localization specialist and we explain the reasoning. And this change is based on user research. And we present this data to localization specialist and say, okay, how can we say this in your language? In the language that you master. This old word has this result and this new word had a better result. So based on the word that had a better result, how can we change in your language, a word that represents this.
Michal: Right. So better result, that's in English. So do you do further testing on other languages after you implement the change?
Willian: We're not AB testing localization yet. It's a plan.
Michal: That's why I asked about optimization. Because I think there is always more that we can learn and always more that we can improve or optimize in our processes. There is so much that we need to take into account when we plan these processes. I think that there's always room for improvement.
Willian: We just implemented this process. It's a baby. And iterating this process and getting data on this process. Like how can we test, how can we measure, what can we do now after we implemented the process? Now that we have a process. This is something that we are still thinking you know, understanding.
Michal: Yeah, definitely. So one last question before we go. You said today that you were manually writing the copy, then you're manually localizing it. So have you considered using any of the machine translation, AI tools, Gen-AI, to kind of streamline your process and make it faster or more efficient?
Willian: In Crowdin I believe we have some MTs, in crowding already. And we make use of a lot of translation memory, but AI is still a, a thing that we're discussing.
Michal: Right. This is like, my... kind of "foot out the door" question. Because, it's a big question. Everyone are... kind of discussing the same thing now, right? And it's out there, and you want to use it, but... you're not really sure how to, and you're not really sure how to implement it, and if it's good in any way. Because if it's not a good fit, then you can do some damage to the process. So it's definitely something that should be approached with caution. But also, I think it's really fascinating to see where it's going to lead everybody.
Is there anything else that you want to say or discussed that we haven't discussed? Because I'd love to hear it too.
Willian: Uh, I mean, Congratulations on putting together that Slack group.
It helped us as professionals to give maturity to the discipline. And I believe it's a topic that has been getting attention lately. Which is really nice because I'm in love with hands-on. So I love creating tutorials, showing how to do stuff and your Slack channel, it'll definitely help me improve and become a better localization specialist here in Quandoo.
Michal: Yeah, that's amazing. It's actually the reason I created this. Because in UX... You're in UX as well, so you know that this community is so good about knowledge sharing, it's so good about bringing people together to kind of share how they do things, and how they set everything up. And they try things and experiment with things and test a lot, and I think that this is a vibe that we should get - and we are getting gradually - in localization. It's definitely something that is going to improve the results, processes, the quality of everything. And I guess, the quality of the experience at the end. So it's definitely worth it.
And for me, that was the main goal for creating that community. But it does depend on people, also, including their own knowledge and sharing their own knowledge and kind of being part of it. And every time someone posts I'm getting really excited to see what they wrote.
Willian: Let's share knowledge and let's help people out there with whatever they're facing or, you know, any challenges they're facing, let's help them.
Michal: Yeah.
This was really, really fascinating. Thank you so much for sharing and thank you so much for being here, for taking the time. I'm actually really, really curious to know where you are going to be in a year from now and where your localization process is going to take you. So hopefully maybe we'll do like a follow up and I'm just going to meet with you again in a year and see how things are going.
Willian: Awesome. Thank you so much Michal.
Michal: Bye bye.
Willian: Bye.
Michal: To everyone listening, if you enjoyed this podcast, if you want to learn more about localization processes and keep up with the next episodes of the Loc Process Pod, make sure you follow Localization Station on LinkedIn or sign up to our email newsletter so I can let you know when a new episode is ready and if you have any feedback, if you have any questions that you would like answered, if there are specific people that you would like to learn about processes from, do let me know.
Send me an email. Send me a message. I would love to hear from you. Thank you so much for taking the time to listen to the lock process pod today. And I hope you have a great day.