Running email marketing in one language is already a challenge. Running it in more than 30 is a completely different level of complexity. From character limits to cultural differences, every detail can change how an email performs.
In this interview, Stas Siachyn shares how he builds and maintains multilingual email across 35+ languages, which tools keep the process under control, and how AI now shapes translation workflows.
Key takeaways
- A clear workflow prevents chaos as you scale. Define who writes, who translates, who reviews, and when each step happens.
- Native speakers and AI work best together. Use AI for speed and rely on native speakers for nuance, legal clarity, and the final polish.
- Strong briefs drive consistency across languages. Character limits, color-coded guidance, and shared translation tables keep CTAs, product names, and key messages on-brand.
- Real tests catch the biggest localization mistakes. Checking links, CTAs, alt text, and language settings inside an actual inbox is the easiest way to avoid errors before launch.
Expert
Since 2017, Stas has worked in email marketing across eCommerce, Web2App, healthtech, and telehealth. He’s recognized for improving deliverability when brands get stuck in spam folders and for helping them generate over $10M in revenue.
His multilingual work spans 35+ languages, including some especially challenging ones, like Hebrew with its right-to-left structure and Japanese with its unique design logic. His experience across so many markets gives a clear, practical look at what it takes to scale email into new regions. If you’re preparing to enter new markets, his lessons will help you avoid the most common mistakes and build a workflow that actually works.
What the workflow looks like and who takes part
Stripo: Who’s involved in your translation workflow? Who does what?
Stas: In one of the projects, I have an in-house customer support team for each language. That helps a lot because they are native speakers, plus they already know our products and customers. It makes translation quality that much better.
Not all projects have enough budget for this approach. In those cases, I hire native speakers on Fiverr. Just make sure they understand your product. Provide a good brief and examples of the same email in English. If translation hits KPIs, come back to the same person. They will learn your product and style over time, and results will grow.
Pro tips:
- A big plus is when the person can do multiple languages; they get better faster.
- Provide a good brief and examples of the same email in English.
- When hiring a native speaker, make sure they understand your product.
If you are a startup, simply use ChatGPT. A good prompt will give you 95% accuracy. Then feed back the translations that worked particularly well. This also gives you a huge advantage: time. You can A/B test almost immediately and it’s very important in the early stages of the brand to identify what the audience likes.
Spend money on native speakers for legal/billing/transactional emails. You want to make sure that these are spotless to avoid confusion.
Stripo: Do you translate emails directly inside your ESP/editor, or through external platforms first?
Stas: I prefer doing it on external platforms. It will usually provide you with more flexibility. The workflow I can advise to you is: ChatGPT > Google Docs > Figma > ESP. But if you want to save time and if your ESP or your email editor has translation functionality, you can definitely use it.
How to keep quality and terminology consistent
Stripo: How do you keep terminology consistent across all languages, especially product names, CTAs, and tone? How do you ensure that updates to terms are reflected everywhere?
Stas: First, you need to know your designs, like:
- How many characters can you use in the image before it breaks the design?
- How many characters for the CTA should be a one-liner?
- How long should the content be to keep the CTR high?
Second, it’s a matter of a good brief. Give the copy for the English version. Comment the character limit on every line. I prefer to color-code my copy: green for a direct translation, while yellow means they can write their own content as long as it makes sense for the email and their market. Main hooks are usually green, like the subject line, header, and CTA. I often leave content blocks, benefits, and stories yellow for native speakers’ interpretation. The more I work with a person (and the more KPIs we hit), the more I leave to their judgment.
Third, I don’t stress about “tone” that much, simply because each language is a different culture. One that I probably don’t understand completely, even after hundreds of tests. When I trust my translator, I get better results every time.
Create a table with a brief and comments in column A, English content in column B, freeze them, and then add a column for each language you need translated. This way, native speakers can reference what others are doing, and get more ideas and insights. Make sure to give access to a specific column only, so they don’t mistakenly change something in another language.
If you choose to translate in ChatGPT, all principles apply. Give a brief and character limit for each line. But you’d need to give a prompt for the content parts to make sure that the LLM takes it as inspiration and writes more specific content for the locale.
Stripo: How do you manage the localization of triggered and transactional emails while keeping tone and logic intact?
Stas: You should definitely invest in proofreading by native speakers for all triggered and transactional emails. There is no way around it. You need a real person to tell you if it makes sense to them and how it makes them feel. Then adjust accordingly. If your budget is tight, bunch them up and send in bulk to save on costs.
Stripo: How do you QA translations before sending?
Stas: Always send yourself an actual email. Open it on multiple devices. Click the links to ensure that you added the corresponding languages in them. Check the translation in DeepL.
Most common mistakes:
- designers forget to change the language in the CTA;
- alternative text for images was not translated;
- the main link and social media links are for the wrong language.
If you have a budget for it, definitely find a native speaker on Fiverr, then send the emails for a proofread.
Send the final design as an actual email to everyone involved in the process. There are many steps where things can go wrong between Figma and the ESP. And since you probably don’t speak 35 languages, always send the final version, not the translation doc or the Figma page.
Stripo: What happens when you tweak a sentence in the English version at the last minute? How do you manage cascading changes?
Stas: ChatGPT is perfect for this kind of task. Give it the updated English version and the current translation. Highlight the changes in the prompt, and you have it done in under 5 minutes.
If you have native speakers in the team, send them a quick message in chat.
But never make it a habit. Translations multiply labor costs. Prepare the English version very carefully. Ideally, prepare the design for the English version as well. It will save you a lot of time and reduce mistakes.
How to enter a new market the right way
Stripo: How do you approach entering a new market?
Stas: I always prepare a “minimal package” for each new market. You just can’t send English. You should consolidate your experience over other markets, and choose emails that work well universally and translate them with ChatGPT.
If the market starts to grow, this is your sign to start doing A/B tests and go beyond your minimal package. How do you know if it’s growing? Mostly, you will be looking at the increase of leads per day. Depending on your product, you will identify how many customers are enough for you to start improving. Could be 100 for some, 10k for others. What’s the “minimal package”? Usually, all transactional emails (without native speaker proofreading), 2-5 abandoned cart emails, some engagement triggers and sunset flow to clean the list efficiently.
Stripo: What should new teams starting their journey in multilingual email do early on to avoid chaos later?
Stas: First, you should decide on the workflow. Is it an English copywriter > English design > native speakers’ translations > localized designs > EPS > test send and proofread? Or is it ChatGPT > Google Docs > Figma > ESP > test send and double check with DeepL? Or something else?
In either case, you need to have an idea in mind of:
- What options do you have available?
- What are your bottlenecks?
- How can you save on labor costs?
For example:
You have native speakers in your team. But you don’t have a dedicated designer and you wait to receive a Figma file. In this case, I’d start the translation process in parallel with design. It will save you days on each email.
And the workflow would be English copywriter > native speakers’ translations > all designs > double check of Figma by native speakers > ESP > test send and proofread.
Once you know your workflow, make sure everyone involved knows it to avoid confusion.
There is no such thing as a perfect workflow. You just have to find what works for your company. You have to listen to feedback and adjust as you go. Don’t worry if it’s chaotic at the start. Localization is a complex problem to solve. Simply focus on improving the process.
Tools that make multilingual work easier
Stripo: What’s your ideal dream tool or platform setup to make all of this easier?
Stas: I kinda mentioned it, but let’s describe it a bit deeper:
- ESP: You need the one that has multilingual functionality. It can be done without it, but trust me, managing 1,000+ templates and 100+ flows in one account is very challenging and unnecessary;
- ChatGPT: If you are serious about your KPIs, you need it to move fast and test a lot. Also helps with ideation and images if you don’t want to wait for your design team. Easily triples the speed at which you can max out our email performance;
- DeepL: Even a free version is good to double-check ChatGPT translations;
- Google Docs: Keep your translations organized;
- Google Sheets: Keep links to your templates/flows organized. You don’t want to rely on your memory to search correctly in ESP. Have everything saved in a table with comments (how many emails are in the flow, were they proofread by a native speaker, do they have active A/B tests, etc.);
- NiftyImages: A niche one, but if you have enough dynamic parameters in your ESP, it can make image localization (and personalization) a much easier process.
Stripo: If you could redesign your multilingual workflow from scratch, what would you keep, and what would you do differently?
Stas: If you want to squeeze the maximum out of your emails, you have to add AI into your multilingual workflow. There is no way around this. It will help you create different tones, angles, and concepts to A/B test. It will save you a lot of money while skyrocketing your KPIs in the shortest amount of time.
Even if you have native speakers in your team, translations usually take a couple of days. With AI help, you can A/B test 5 different versions of each language already. Then, when you have winning angles, you can polish them with a native speaker. But still, A/B test human translations with the AI version. You would be surprised how often highly tested AI content is going to win over a native translator version. And I would always keep a human touch on transactional emails. Marketing can be weird as long as it works, and transactional has to make sense.
Wrapping up
Scaling email marketing across multiple languages is a long-term effort, but it becomes manageable with the right structure. Clear briefs, a reliable workflow, and fast testing help teams avoid chaos, while native speakers handle the nuance that AI can’t replace. Whether you’re entering your first new market or managing dozens of languages, the principles discussed above make multilingual email easier, faster, and more consistent, without lowering quality.
0 comments