Lokalise 101: Unlock Seamless Localization

Full transcript

Welcome and introductions

Thomas Rickard: Welcome, everyone, and thank you so much for joining this first session in a long line of the Lokalise series. I'm Thomas, a Senior Customer Success Manager at Lokalise — I've been here for just over three years. We have a packed agenda today: we're going to explore how Lokalise helps streamline the translation and localization process, walking through everything from setting up your first project to managing tasks and understanding translation types. By the end of this session, the idea is for you to have a solid foundation to start actually using Lokalise.

The intended audience for today's workshop is somebody who is new to Lokalise, or somebody who hasn't used Lokalise for some time and would like a refresher. It's intended to be an engaging session, so all questions are welcome — please add them to the Q&A box. The session combines a slide deck presentation with a live demo: the presentation is segmented into three chapters, and after each chapter I'll log in to Lokalise and visually recap in a demo environment what we've just covered. I have two wonderful colleagues, Brittany and Mikaela, helping me with the Q&A. If there are any outstanding questions or we run out of time, we will of course get back to you afterwards. You'll also receive a copy of today's materials and the recording.

Before I get started, let's run a quick poll so I can get an understanding of the room and your knowledge of Lokalise. Okay — the majority of people in the room today have never used Lokalise, so this is absolutely perfect. And for everyone else, this is a great opportunity for a refresher; things may have changed since the last time, and it's a chance to ask questions.

Brittany: Reading off what people are hoping to learn: to be able to use it for a new project; just generally how to use it; "my team has used it for months, but I would like a 101"; solving some doubts about using Lokalise; how to use it for workflow management and translation processes; getting started with the basics; a better overall understanding and an intro to the tool. And someone says they hope it's not a sales pitch — we promise not to do that. It sounds like the general consensus is a basic understanding of how to use it.

Thomas Rickard: Brilliant, thank you. So let's begin with the why and set the scene for Lokalise. A common challenge many teams face is getting high-quality translation — often, inefficiencies within the process make localization a daunting task. Today I'll focus on how Lokalise can eliminate those challenges, streamline the translation process, and make it easier to achieve results from the start. Collaboration matters too: we'll also look at how to add colleagues and collaborate on the workspace. Without the right tools and experience, localization can become a bottleneck for many organizations. The goal today is to set you up for success — showing you how to use Lokalise effectively from the beginning so that you don't just get translations done, but do them the right way: maximizing efficiency, quality, and collaboration, and avoiding common pitfalls.

Creating your account and understanding roles

Thomas Rickard: We'll begin at the beginning. First, you'll need to create an account, and this happens even if you're added to an existing workspace. If somebody adds you to a workspace, you'll receive an email; when you click the link, it prompts you to fill in some details about yourself and your role. The idea is that the platform can immediately configure itself to your needs and help you hit the ground running without wading through tons of irrelevant menus and settings.

Once your account is set up, the next step is to add team members to your workspace and then to the relevant projects. In Lokalise, you can invite team members and assign them specific roles — developers, reviewers, content creators, translators — each with different access levels. This ensures that only the right people can make changes to your project. For example, a developer has full access to keys and files, while a translator can only work with assigned languages. This flexibility is key to maintaining control over your project while allowing collaboration between different contributors. For those of you who have used Lokalise before: we released much more granular permissions about five months ago, and you can even create custom roles.

There are five core roles, each with a specific permission set. The manager has full access to everything, including project settings, integrations, and user management. The developer can upload and download files, manage keys, and integrate with external tools. Content creators can manage translation keys, add screenshots, and add languages, but don't have access to project-wide settings. A reviewer can review and approve translations — the goal for reviewers is to ensure quality control. Lastly, translators work only with assigned languages in the most restricted mode: a translator has a very limited view of Lokalise, and their role is specifically to translate the languages they've been given permission to view and edit, and nothing else.

Creating your first project

Thomas Rickard: When creating a project, you'll have three different project types to choose from, and depending on the kind of content you're translating, each has slightly different functionality. The first is web and mobile — ideal for translation files such as JSON, YAML, or XML, and the right choice if you plan to use any of our out-of-the-box integrations like GitHub, WordPress, or Figma. Next is marketing and support — use this if you're dealing with content from services like Marketo, Zendesk, or Salesforce. Lastly, ad hoc documents — perfect for translating documents like DOCX, PPTX, HTML, and lots of other text-based formats. Once you've selected your project type, Lokalise guides you through the setup, and you'll be ready to add your content and start collaborating.

Now let me log in and quickly go through what we've covered. After entering your details, you'll see your first login screen — please don't let it scare you. Lokalise is a project-based platform, so what you see are all the projects I have access to or have created; it's drag and drop, and you can arrange them however you like.

To add team members to the workspace, click on your picture in the bottom left (a generic symbol when you first log in) and go to team settings. Click to add a team member, enter their email address, and add them to projects. Then give them a role — in this case a translator, which will come in handy later when we assign work. Next you set editable and readable languages: I want this translator to edit French and German only, while being able to read the other languages — although I can limit that view too.

To create a project, click new project and pick from the three project types — web and mobile for today. Give it a name and set the base language: that's the language you're uploading content in, English in this instance. From here you add all the languages you need, like French and German. Importantly, this can be changed at any point — you can add and remove languages later, so you don't need to add twenty-five languages right now if you only need one or two; as you grow and expand into new regions, you can always add more. Then click create project.

From there it asks you to upload files, or you can import directly from our out-of-the-box integrations under the apps section. Anything out of the box means the integration is ready-made — you just grant permission to connect. For Figma, for example, you log in to your Figma instance and allow access; it takes quite literally twenty or thirty seconds. For today I'll upload a simple file. On the right-hand side you'll notice one option is prefilled: replace line break. That's just coding language for where a word ends and a new one begins — it's prepopulated, so you don't need to worry about it. Click import, and the file is uploaded into the editor.

One useful lesson from a mistake I just made: I added my teammate to the wrong project. To fix it, open the correct project, click the contributors tab, and add them there — everything can be edited after the fact.

Inside the editor

Thomas Rickard: With a file uploaded, we're now in the editor tab — essentially where the entire translation process takes place, and where most people spend most of their time. Let's look at the important functions on screen. First, filters: they let you filter keys by different criteria, like untranslated, platform, file name, or import date, and you can create your own custom filters. Beside that is the sorting option — sort by last added is the default, but you can also sort alphabetically and by other attributes.

Next is the project overview, which gives you the total number of translation keys and the total number of words in the project. Then the key name, along with the platform that key is associated with — for example, a little Chrome icon indicates web. Within one project you can have multiple platforms — Android, iOS, web — all living together; you don't need separate projects for each. The question I get asked all the time is what best practice is, and it really depends on the preferred method and makeup of your team: some organizations keep one platform per project, others keep everything in one project. There are little nuances when it comes to retranslating the same content, but multiple platforms can live in the same project.

Next, a simple but super useful option: remember my place. In a project with lots of keys — this one has 462 — toggling this means the next time you log in to Lokalise, it remembers where you finished. If I was on key 231 when I logged out, it brings me straight back to that key. Not something that will rock your world, but it certainly makes life easier.

Then there's the more option, a drop-down menu with project settings and automations. Next, add key: at any point you can manually add a single key instead of using a file import or an app integration. There's also a search bar for finding specific translation keys or phrases in a large project. And lastly, status indicators: icons that provide a quick visual indication of the current status of each string — four icons plus a warning sign, which we'll walk through in the demo.

Within the editor you can also change the view. The bilingual view is designed for projects involving two languages only, showing the source and target language side by side — making it easier to compare translations and keep consistency. The multilingual view is ideal for projects with two or more languages: you see multiple target languages at once, allowing a broader perspective and easier management across several languages. Switching between them is a single click.

Translation types and translation memory

Thomas Rickard: Before going further, a quick poll: how are you typically getting your content translated right now? A lot of in-house translators, some LSPs. So — in Lokalise there are several different translation types available, and it's important to understand how each works so you can use the platform effectively.

First, human translation: human translators are the most accurate — they're context-aware and understand the nuances of language and culture. Second, machine translation: it uses algorithms and models to translate content, and while it's faster than humans, it lacks the context and nuances a human might catch. A great example is European Spanish versus Latin American Spanish — machine translation will translate, but it may not pick up the slight differences in nuance between the two. Third, AI translation: AI translates with context — you create context via a style guide or description notes, which helps make it more accurate — and it's also faster than humans. Understanding these types is critical because each has its own strengths.

Then there's translation memory — a system that stores your previous translations and suggests them when similar content appears again. It saves time, ensures consistency across translations, reduces costs, and is especially useful for repetitive content, automatically providing suggestions based on previous translations.

Translation memory provides suggestions in four categories: an exact match (a perfect match with a previous translation), a context match (which shows as 101% — a match where the surrounding context also matches), a fuzzy match (partial — something similar but not identical), and no match (no previous translation found). Importantly, you own your translation memory. If you're coming to Lokalise from a previous tool, or even from Excel sheets, you can upload what you have. If you don't have one, Lokalise creates one for you as you translate — and you own it and can download it at any point and take it with you. It ensures consistency, and commercially it saves money long term, because you don't need to retranslate the same thing twice.

In the more settings, you can set your translation memory priority. If you have more than one translation memory, you control how suggestions are made by setting the priority order. A great use case is large organizations with a siloed approach: the marketing team might have its own translation memory and the legal team another — legal documentation has to comply with various regulations, while marketing might be a bit more fun and less formal. With a rule in place, Lokalise searches your translation memories in priority order for an exact match before translating.

Back in the editor demo: on the left are your keys, on the right your strings with all available languages, and you can toggle between the multilingual and bilingual views. You can add or remove languages here at any point, and you can even change the base language with a new file import or by manually changing it — most functionality in Lokalise can be edited as you go, so you don't need to create new projects.

On the right-hand side of each string there's a comments section — as people translate, conversations about a particular translation take place there. Next, the glass symbol: mark as reviewed, which is really important for QA reviewers. It signals that eyes have been on this string, it's good to go for export, and everything is confirmed. The wavy lines are the opposite — unverified — meaning something may be wrong, perhaps grammatical errors or spelling mistakes. There's also an amber warning sign: when highlighted, it generally means something is off and the string needs double-checking. And remember, you can filter between these statuses.

Lastly, the history option: at any point you can check the history of a particular string. Thinking about continuous localization — you don't need to create a new project every time; you upload new versions and existing strings simply update with the new values. At all points you'll see the difference: what the translation was, what it is now, who made the change, when, and where the import took place. There's also a restore option — if an import today introduced a mistake, you can restore the previous version. Nothing will be lost or broken.

Clicking into a target language shows the suggestion panels: translate with AI at the top (which asks for context), then translation memory — here a 100% match showing the previous translation and when it was made — and then the machine translation options. You don't need to click into each string individually: automations can make all of this happen automatically, but it's important to know what each option means.

Machine translation, AI, and working with translation providers

Thomas Rickard: Let's dive into the non-human translation options in Lokalise and their nuanced differences. Machine translation allows automatic translations through third-party providers — Google, DeepL, and Microsoft — and each provider has different capabilities. There are inline suggestions, which we saw when clicking within a string, and there are automations you can set up, such as pre-translation rules (not every vendor is available for every automation — it depends on what they allow). You can automate all of these actions: for example, on import your translation memory translates first, then machine translation fills the gaps. You set the parameters; it's entirely up to you.

Lokalise AI is powered by OpenAI (ChatGPT). The question I get asked constantly is: is AI meant to replace language service providers, or humans? Absolutely not — it's another layer on top. There are two workflows I see with the customers I work with today: either AI does the first round of translations and human reviewers double-check, or the opposite — humans translate and AI reviews their work. It's meant to be an added layer of extra value. That said, there are some organizations using Lokalise today who only use AI and are very happy with the quality output, so it's definitely something to consider — you can run proofs of concept with AI in Lokalise to see what fits your needs, particularly if you're working with professional translators where contracts and costs vary. With AI, costs are a little more upfront.

It's also important to mention that Lokalise supports working with language service providers and freelancers for more specialized translations. For anyone not from a localization background: LSPs are companies that offer translation services, often with teams of translators specializing in different industries or languages; freelancers are independent translators you can hire for specific projects. You can absolutely use both on Lokalise. When working with either, be clear and upfront: define your timelines, the scope, what's expected, and when. We make that easy with task assignments, due dates, and custom statuses so everyone knows where things stand.

And think about collaboration with context. Professional translators will ask for context, so you can add screenshots and comments to the actual keys so they can see what they're translating, and use the in-editor chat to clarify things. It's a game changer compared to working with static spreadsheets — live conversations connected to strings, questions answered in place, and content visible in context. Content imported from Figma comes with its frames and screenshots attached, and file formats like HTML give translators an on-screen preview of exactly what they're translating for in-context review.

Tasks: assigning and tracking work

Thomas Rickard: Let's tie it all together. We've done our file imports and looked at the editing options — now let's look at actually giving a translator or an in-house native speaker work to do. How do we tell them there's work, instead of just sending an email, and how do we track progress? This is where tasks come in.

Tasks are used to assign specific pieces of content for translation or review. You can create tasks for translators, reviewers, or AI — you need to tell AI to translate, and this is how you do it in bulk. You select the task type, choose AI or a human already on your workspace, give it context if needed, and add due dates and descriptions. There are also more advanced options: you can auto-close tasks when translations are completed, and you can chain tasks — open task A for a translator to translate and task B for a reviewer, and when the translation task closes, the review task automatically opens. That's really important for QA.

Once created, task management in Lokalise lets you monitor and track progress — for translations, reviews, and AI translations alike. You can see what's completed, in progress, or pending (a chained task that hasn't been activated yet shows as pending), all within the task dashboard. Widgets and on-screen pop-ups let you know when things happen, and assignees get emails, due-date prompts, and any descriptions and context you've added.

In the demo: you can create a task from two places — directly from the editor or from the task dashboard; generally you'd do it from the editor. Select individual keys or select all (I have seven here). Create task is the first option in the menu — there are roughly twenty or thirty different options in that menu, but let's not confuse everyone today. The task types are: a translation task for native speakers or professional translators; translate with AI; a review task (again, you can create chained tasks); and lastly LQA with AI, which runs a quality assurance check on existing translations. Thinking about that workflow where humans translate first and you want AI to double-check: AI produces a report of its findings — for example, key number one has grammar issues or misspellings.

For today, a translation task: name it, add a description, optionally a due date. The source language defaults to your base language, English, and all the project's languages are targets — remove any you don't need. Notice the assignees: me by default, plus the teammate we added earlier who has French and German permissions, so they're automatically added. If you had more people with those language permissions, they'd appear here and you'd select who to assign. Then simply create the task. It redirects to the task workspace: our new task shows French, sixteen words, no progress yet — as the assignee translates, the progress bar moves.

Finally, data. In the analysis section, the initial analysis is available right away; after work is done you can view a final report and a detailed translation report. Depending on your contract and how you use Lokalise, there's also Lokalise Analytics, which gives a full breakdown of absolutely everything happening in your workspace per language, plus a task dashboard view from start to finish. You can also see post-edit rates for each language — important to check, because post-edit rates are a great indicator of quality. They should be as low as possible; on average, fourteen to fifteen percent is very acceptable.

Resources and getting help

Thomas Rickard: In terms of where to go for help, we have an abundance of information available. There's Lokalise Academy — it asks you to sign up, and you won't be bombarded with emails — with different courses depending on your job title. It's snackable content: not ninety-minute webinars, but very short lessons depending on what you're trying to do. We also have our user guides and FAQs — questions like the one about keyboard shortcuts are answered there. Please keep the questions coming, and thank you all for joining.

Q&A

Q: Can I link more than one Lokalise project to the same Figma file?
Correct — you can't, because the connection is to one instance. A Lokalise project is connected to one Figma instance. However, you can have multiple Figma projects within that one Lokalise project.

Q: Does the tool support SDLXLIFF files?
I need to double-check — I'm not entirely sure offhand. We'll follow up with you after the session.

Q: Is there any way to customize keyboard shortcuts? It would be great to press Ctrl+Enter to confirm a segment and jump to the next one without clicking.
We do have keyboard shortcuts, and there's an entire article covering what shortcuts do what, when, and where — you'll find it in the resources shared at the end of the presentation.