![\"Get](\"https:\/\/no-cache.hubspot.com\/cta\/default\/53\/3e25e192-30c3-40c1-a7da-a4d054c9e157.png\")
<\/a><\/p>\nThis wasn\u2019t my first time navigating global nuance. With an MBA in International Business and experience working on a global consulting project in Portugal, I\u2019d already seen how messages land differently depending on culture, tone, and language. But this was my first time applying that lens to AI content generation in my MarTech AI role at HubSpot \u2014 and it was more complex than expected.<\/p>\n
We already had an AI-generated email prompt that worked well in English\u2014conversational, friendly, and context-aware. The challenge? Making it work in Spanish and French without sounding robotic, clumsy, or culturally off-base.<\/p>\n
Sounds easy. It wasn\u2019t.<\/p>\n
The Hidden Complexity of “Just Localizing”<\/strong><\/h2>\nWhat we were really doing was asking an AI model \u2014 trained predominantly in English \u2014 to speak other languages as naturally as a native marketer would<\/em>.<\/p>\nOur first attempts fell flat.<\/p>\n
Example (original AI output in Spanish):<\/strong><\/p>\nHere\u2019s what we aimed for in English:<\/p>\n
\u201cI saw you were scoping around the platform and that you were interested in speaking with us. Would you like to meet on one of the following days?\u201d<\/em><\/p>\nThis is the original output in Spanish:<\/p>\n
\u201cEstuve revisando tus interacciones en nuestra plataforma y quer\u00eda ofrecerme como tu punto de contacto.\u201d<\/em><\/p>\nIn English, it translates to:<\/p>\n
\u201cI reviewed your activity and wanted to become your point of contact.\u201d<\/em><\/p>\nWhile grammatically correct, this sounded invasive<\/em> in Spanish \u2014 like we were watching the user too closely. It didn\u2019t feel natural. One reviewer called it \u201ccreepy.\u201d<\/p>\nHere\u2019s another example:<\/p>\n
\n- Original English intent: <\/strong>\u201cI noticed you\u2019ve been exploring our platform and expressed interest in connecting with us.\u201d<\/em><\/li>\n
- Original Spanish output:<\/strong> \u201cMe pareci\u00f3 interesante tu inter\u00e9s en nuestros servicios.\u201d<\/em><\/li>\n
- Translation in English:<\/strong> \u201cI found your interest in our services interesting.\u201d<\/em><\/li>\n<\/ul>\n
Again, it\u2019s technically accurate, but it\u2019s redundant and robotic. It\u2019s the kind of phrasing that makes a reader stop and go, \u201cDid a bot write this?\u201d<\/p>\n
The takeaway: <\/strong>Even when the translation is accurate, the tone<\/em> can be off. And tone is everything in marketing.<\/p>\nThe Shift from Translation to Language-aware Prompt Design<\/strong><\/h2>\nAt this point, I realized we needed more than AI outputs \u2014 we needed a system<\/strong> for guiding the AI to think like a multilingual marketer.<\/p>\nI built a language-portable prompt framework<\/strong> \u2014 a structured prompt that could adapt across languages while respecting each one’s unique grammar, tone, and cultural context.<\/p>\nHere\u2019s What Changed<\/strong><\/h3>\nInstead of one static prompt, I broke the logic into variables:<\/p>\n
\n- : Target language (e.g., Spanish, French, German)<\/li>\n
- : Pronoun and tone level (\u201ctu\u201d vs. \u201custed\u201d, \u201cvous\u201d vs. \u201ctu\u201d)<\/li>\n
- : Inbox-friendly, conversational, professional<\/li>\n
- : Direct vs. suggestive phrasing<\/li>\n
- : Enforced where grammar allowed<\/li>\n<\/ul>\n
We also added clear, language-specific rules.<\/p>\n
Example (Spanish):<\/strong><\/p>\n\n- Use t\u00fa<\/em> consistently, never usted<\/em> (too formal for our brand)<\/li>\n
- Avoid gendered adjectives like interesado\/interesada<\/em> when possible
\u2705 \u201cMostraste inter\u00e9s en \u2026 \u201d
<\/em> \u274c \u201cEstuviste interesado en \u2026 \u201d<\/em><\/li>\n<\/ul>\nExample (French):<\/strong><\/p>\n\n- Always use vous<\/em>, not tu<\/em>, in B2B messages<\/li>\n
- Avoid ambiguous endings like int\u00e9ress\u00e9(e)
<\/em> \u2705 \u201cVous avez montr\u00e9 de l\u2019int\u00e9r\u00eat \u2026 \u201d
<\/em> \u274c \u201cTu t\u2019\u00e9tais int\u00e9ress\u00e9(e) \u2026 \u201d<\/em><\/li>\n<\/ul>\nWhy This Shift Mattered<\/strong><\/h3>\nIn English, a friendly CTA might look like:<\/p>\n
\u201cWould you be available for a brief conversation on one of the following days?\u201d<\/em><\/p>\nWe tried directly translating it into Spanish:<\/p>\n
\u201c\u00bfQuieres agendar 15 minutos para hablar sobre lo que est\u00e1s buscando?\u201d<\/em><\/p>\nIt was grammatically correct, but it sounded too casual<\/em> and unprofessional in a B2B context. Not pushy, just slightly off-tone.<\/p>\nSo, we reworded it to be friendly but formal:<\/p>\n
\u201cSi te parece bien, podemos agendar una conversaci\u00f3n breve esta semana.\u201d<\/em><\/p>\nThis translates to:<\/p>\n
\u201cIf it works for you, we can schedule a short chat this week.\u201d<\/em><\/p>\nHere\u2019s another example in French:<\/p>\n
\n- Original output<\/strong>: <\/strong>\u201cSouhaitez-vous prendre rendez-vous pour en discuter ?\u201d
<\/em> (\u201cWould you like to schedule a meeting to discuss this?\u201d<\/em>)<\/li>\n- New version: <\/strong>\u201cAuriez-vous 20 minutes pour voir comment HubSpot pourrait concr\u00e8tement vous aider?\u201d<\/em> (\u201cWould you have 20 minutes to see how HubSpot could practically support you?\u201d<\/em>)<\/li>\n<\/ul>\n
The second version adds value to the CTA. Not just time \u2014 but purpose.<\/p>\n
Backing It Up With a Stakeholder Questionnaire<\/strong><\/h2>\nLocalization isn\u2019t just a linguistic issue \u2014 it\u2019s a business alignment issue.<\/p>\n
To get it right, I created a simple stakeholder intake doc and shared it with marketing ops, regional marketers, and content leads. The goal was to align early on tone, content boundaries, and regional sensitivities.<\/p>\n
These are some of the questions I asked:<\/p>\n
\n- What level of formality is appropriate in your market?<\/li>\n
- Should we avoid gendered terms?<\/li>\n
- Can we reference the user\u2019s company or product usage?<\/li>\n
- How direct should we be in asking for action?<\/li>\n
- Are there idioms, cultural references, or phrasings we should avoid?<\/li>\n<\/ul>\n
We got some pretty interesting insights.<\/p>\n
For example, in some regions, stakeholders preferred not<\/em> to reference the recipient\u2019s company type in the copy, even though that was common in English (e.g., \u201cI saw that you help startups with HR\u201d).<\/p>\nThe localized alternative became more general:<\/p>\n
\u201cEntiendo que est\u00e1n buscando formas de mejorar sus procesos internos.\u201d <\/em>(\u201cI understand you\u2019re looking to improve internal processes.\u201d<\/em>)<\/p>\nThe results of this survey helped create clarity between content, ops, and regional marketing teams \u2014 and dramatically reduced our revision cycles.<\/p>\n
The Final Product: Human-sounding Emails at Scale<\/strong><\/h2>\nWith the updated prompt and intake framework, the new outputs were instantly better.<\/p>\n
Before:<\/strong><\/p>\n\n- Original output: <\/strong>\u201cHola [FirstName], soy Mar\u00eda de HubSpot. He visto que has navegado nuestra plataforma y parece que te interesa nuestro producto.\u201d<\/em><\/li>\n
- English translation: <\/strong>\u201cHi [FirstName], I\u2019m Mar\u00eda from HubSpot. I saw you\u2019ve browsed our platform and it seems you\u2019re interested in our product.\u201d<\/em><\/li>\n<\/ul>\n
After:<\/strong><\/p>\n\n- Original output: <\/strong>\u201cSoy Mar\u00eda de HubSpot. Vi que estuviste explorando la plataforma y que quer\u00edas saber m\u00e1s sobre c\u00f3mo podemos apoyar tu negocio.\u201d<\/em><\/li>\n
- English translation:<\/strong> \u201cI\u2019m Mar\u00eda from HubSpot. I saw you were exploring the platform and wanted to learn more about how we can support your business.\u201d<\/em><\/li>\n<\/ul>\n
And stakeholders responded positively:<\/p>\n
\n- \u201cThis finally sounds like someone from our team wrote it.\u201d<\/li>\n
- \u201cPerfect tone \u2014 natural and local.\u201d<\/li>\n
- \u201cNo gender errors or weird formalities. We can actually use this.\u201d<\/li>\n<\/ul>\n
Even better, we didn\u2019t need to write separate prompts for every campaign. The same core framework now powers AI-generated messages in multiple languages \u2014 with consistent quality.<\/p>\n
Takeaways for Marketers<\/strong><\/h2>\nWhether you\u2019re working on AI copy, global ads, or multilingual content, here\u2019s what I learned:<\/p>\n
1. Don\u2019t just translate \u2014 localize for intent.<\/strong><\/h3>\nLiteral translations will get you \u201ctechnically correct\u201d content. But only localization will make it land.<\/p>\n
2. Use prompts like creative briefs.<\/strong><\/h3>\nInclude tone, formality, CTA style, gender neutrality, and other language rules as variables. Don\u2019t leave nuance to chance.<\/p>\n
3. Build language-aware templates.<\/strong><\/h3>\nLanguages behave differently. Plan for things like verb conjugations, pluralization, and sentence rhythm upfront.<\/p>\n
4. Get feedback early.<\/strong><\/h3>\nUse a stakeholder intake doc before generation, not after. You\u2019ll avoid rework and misalignment later on.<\/p>\n
5. Aim for a real, human tone.<\/strong><\/h3>\nIf your AI output doesn\u2019t feel like something you<\/em> would write to a customer, it won\u2019t convert. Read it aloud. Would you hit send?<\/p>\nAI localization is a marketing skill now.<\/strong><\/h2>\nThis project taught me something that has stuck with me since: The future of global marketing<\/strong> isn\u2019t just about scaling content \u2014 it\u2019s about scaling context.<\/strong><\/p>\nThe companies that succeed with AI won\u2019t be the ones who generate the most content. They\u2019ll be the ones who generate the most resonant<\/em> content because they know how to prompt for it. And that starts with understanding the languages your customers speak \u2014 in more ways than one.<\/p>\n![\"\"](\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\")
<\/p>\n","protected":false},"excerpt":{"rendered":"
Earlier this year, I was handed an AI-generated content project with a deceptively simple goal: adapt email messages for international audiences. This wasn\u2019t my first time navigating global nuance. With an MBA in International Business and experience working on a global consulting project in Portugal, I\u2019d already seen how messages land differently depending on culture, […]<\/p>\n","protected":false},"author":1,"featured_media":1180,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[16],"tags":[],"class_list":["post-1178","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-featured"],"_links":{"self":[{"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/posts\/1178","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/comments?post=1178"}],"version-history":[{"count":1,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/posts\/1178\/revisions"}],"predecessor-version":[{"id":1179,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/posts\/1178\/revisions\/1179"}],"wp:featuredmedia":[{"embeddable":true,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/media\/1180"}],"wp:attachment":[{"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/media?parent=1178"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/categories?post=1178"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/tags?post=1178"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
Our first attempts fell flat.<\/p>\n
Example (original AI output in Spanish):<\/strong><\/p>\n Here\u2019s what we aimed for in English:<\/p>\n \u201cI saw you were scoping around the platform and that you were interested in speaking with us. Would you like to meet on one of the following days?\u201d<\/em><\/p>\n This is the original output in Spanish:<\/p>\n \u201cEstuve revisando tus interacciones en nuestra plataforma y quer\u00eda ofrecerme como tu punto de contacto.\u201d<\/em><\/p>\n In English, it translates to:<\/p>\n \u201cI reviewed your activity and wanted to become your point of contact.\u201d<\/em><\/p>\n While grammatically correct, this sounded invasive<\/em> in Spanish \u2014 like we were watching the user too closely. It didn\u2019t feel natural. One reviewer called it \u201ccreepy.\u201d<\/p>\n Here\u2019s another example:<\/p>\n Again, it\u2019s technically accurate, but it\u2019s redundant and robotic. It\u2019s the kind of phrasing that makes a reader stop and go, \u201cDid a bot write this?\u201d<\/p>\n The takeaway: <\/strong>Even when the translation is accurate, the tone<\/em> can be off. And tone is everything in marketing.<\/p>\n At this point, I realized we needed more than AI outputs \u2014 we needed a system<\/strong> for guiding the AI to think like a multilingual marketer.<\/p>\n I built a language-portable prompt framework<\/strong> \u2014 a structured prompt that could adapt across languages while respecting each one’s unique grammar, tone, and cultural context.<\/p>\n Instead of one static prompt, I broke the logic into variables:<\/p>\n We also added clear, language-specific rules.<\/p>\n Example (Spanish):<\/strong><\/p>\n Example (French):<\/strong><\/p>\n In English, a friendly CTA might look like:<\/p>\n \u201cWould you be available for a brief conversation on one of the following days?\u201d<\/em><\/p>\n We tried directly translating it into Spanish:<\/p>\n \u201c\u00bfQuieres agendar 15 minutos para hablar sobre lo que est\u00e1s buscando?\u201d<\/em><\/p>\n It was grammatically correct, but it sounded too casual<\/em> and unprofessional in a B2B context. Not pushy, just slightly off-tone.<\/p>\n So, we reworded it to be friendly but formal:<\/p>\n \u201cSi te parece bien, podemos agendar una conversaci\u00f3n breve esta semana.\u201d<\/em><\/p>\n This translates to:<\/p>\n \u201cIf it works for you, we can schedule a short chat this week.\u201d<\/em><\/p>\n Here\u2019s another example in French:<\/p>\n The second version adds value to the CTA. Not just time \u2014 but purpose.<\/p>\n Localization isn\u2019t just a linguistic issue \u2014 it\u2019s a business alignment issue.<\/p>\n To get it right, I created a simple stakeholder intake doc and shared it with marketing ops, regional marketers, and content leads. The goal was to align early on tone, content boundaries, and regional sensitivities.<\/p>\n These are some of the questions I asked:<\/p>\n We got some pretty interesting insights.<\/p>\n For example, in some regions, stakeholders preferred not<\/em> to reference the recipient\u2019s company type in the copy, even though that was common in English (e.g., \u201cI saw that you help startups with HR\u201d).<\/p>\n The localized alternative became more general:<\/p>\n \u201cEntiendo que est\u00e1n buscando formas de mejorar sus procesos internos.\u201d <\/em>(\u201cI understand you\u2019re looking to improve internal processes.\u201d<\/em>)<\/p>\n The results of this survey helped create clarity between content, ops, and regional marketing teams \u2014 and dramatically reduced our revision cycles.<\/p>\n With the updated prompt and intake framework, the new outputs were instantly better.<\/p>\n Before:<\/strong><\/p>\n After:<\/strong><\/p>\n And stakeholders responded positively:<\/p>\n Even better, we didn\u2019t need to write separate prompts for every campaign. The same core framework now powers AI-generated messages in multiple languages \u2014 with consistent quality.<\/p>\n Whether you\u2019re working on AI copy, global ads, or multilingual content, here\u2019s what I learned:<\/p>\n Literal translations will get you \u201ctechnically correct\u201d content. But only localization will make it land.<\/p>\n Include tone, formality, CTA style, gender neutrality, and other language rules as variables. Don\u2019t leave nuance to chance.<\/p>\n Languages behave differently. Plan for things like verb conjugations, pluralization, and sentence rhythm upfront.<\/p>\n Use a stakeholder intake doc before generation, not after. You\u2019ll avoid rework and misalignment later on.<\/p>\n If your AI output doesn\u2019t feel like something you<\/em> would write to a customer, it won\u2019t convert. Read it aloud. Would you hit send?<\/p>\n This project taught me something that has stuck with me since: The future of global marketing<\/strong> isn\u2019t just about scaling content \u2014 it\u2019s about scaling context.<\/strong><\/p>\n The companies that succeed with AI won\u2019t be the ones who generate the most content. They\u2019ll be the ones who generate the most resonant<\/em> content because they know how to prompt for it. And that starts with understanding the languages your customers speak \u2014 in more ways than one.<\/p>\n Earlier this year, I was handed an AI-generated content project with a deceptively simple goal: adapt email messages for international audiences. This wasn\u2019t my first time navigating global nuance. With an MBA in International Business and experience working on a global consulting project in Portugal, I\u2019d already seen how messages land differently depending on culture, […]<\/p>\n","protected":false},"author":1,"featured_media":1180,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[16],"tags":[],"class_list":["post-1178","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-featured"],"_links":{"self":[{"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/posts\/1178","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/comments?post=1178"}],"version-history":[{"count":1,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/posts\/1178\/revisions"}],"predecessor-version":[{"id":1179,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/posts\/1178\/revisions\/1179"}],"wp:featuredmedia":[{"embeddable":true,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/media\/1180"}],"wp:attachment":[{"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/media?parent=1178"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/categories?post=1178"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.coclea.org\/index.php\/wp-json\/wp\/v2\/tags?post=1178"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}\n
The Shift from Translation to Language-aware Prompt Design<\/strong><\/h2>\n
Here\u2019s What Changed<\/strong><\/h3>\n
\n
\n
\u2705 \u201cMostraste inter\u00e9s en \u2026 \u201d
<\/em> \u274c \u201cEstuviste interesado en \u2026 \u201d<\/em><\/li>\n<\/ul>\n\n
<\/em> \u2705 \u201cVous avez montr\u00e9 de l\u2019int\u00e9r\u00eat \u2026 \u201d
<\/em> \u274c \u201cTu t\u2019\u00e9tais int\u00e9ress\u00e9(e) \u2026 \u201d<\/em><\/li>\n<\/ul>\nWhy This Shift Mattered<\/strong><\/h3>\n
\n
<\/em> (\u201cWould you like to schedule a meeting to discuss this?\u201d<\/em>)<\/li>\nBacking It Up With a Stakeholder Questionnaire<\/strong><\/h2>\n
\n
The Final Product: Human-sounding Emails at Scale<\/strong><\/h2>\n
\n
\n
\n
Takeaways for Marketers<\/strong><\/h2>\n
1. Don\u2019t just translate \u2014 localize for intent.<\/strong><\/h3>\n
2. Use prompts like creative briefs.<\/strong><\/h3>\n
3. Build language-aware templates.<\/strong><\/h3>\n
4. Get feedback early.<\/strong><\/h3>\n
5. Aim for a real, human tone.<\/strong><\/h3>\n
AI localization is a marketing skill now.<\/strong><\/h2>\n
![\"\"](\"https:\/\/track.hubspot.com\/__ptq.gif?a=53&k=14&r=https%3A%2F%2Fblog.hubspot.com%2Fmarketing%2Flocalize-ai-generated-emails&bu=https%253A%252F%252Fblog.hubspot.com%252Fmarketing&bvt=rss\")
<\/p>\n","protected":false},"excerpt":{"rendered":"