Global companies collect feedback in dozens of languages. Too often, the survey is translated word for word and the answers are translated back the same way. The result is data that looks complete and still misses what customers meant.
Multilingual Feedback Is Now Normal
A major survey platform's 2025 State of Surveys report found that 21% of surveys in 2024 included at least one translation, up from 9% in 2014. More teams are listening across languages. Fewer are doing it well.
Customers notice bad translation
A recent Liferay survey of 800 US consumers found that more than half of multilingual shoppers have walked away from a purchase because an experience seemed poorly localized. Yotam Levy of Liferay said of companies, “The problem is they don't necessarily do a great job of localization.”
Surveys with at least one translation
Source: 2025 State of Surveys report. Figures as cited in this article
Where Multi-Language VoC Breaks
- Survey questions lose meaning when idioms don't translate.
- Rating scales carry different weight across cultures.
- Machine translation of answers strips out sarcasm and tone.
- Teams compare scores across countries as if they meant the same thing.
Each of these problems is fixable once you know to look for it.
Scales vary by culture
In some cultures, customers rarely give top scores. In others, a middle score signals real unhappiness. Comparing raw NPS across regions without that context leads to wrong conclusions. Look at the spread of scores in each market before you set targets.
Fix the Survey Side
Translate each question for meaning. Have a native speaker review each version and test it with a few local customers. Keep the intent of every question identical even when the wording changes. Pay close attention to scale labels. Phrases like “very satisfied” don't always have clean equivalents, and small shifts change how people score.
Localize the whole experience
Dates and currency should follow local conventions. Customers expect a survey to feel like it was written for them. A survey that feels foreign gets fewer and weaker answers.
Free review snapshot
See what your customers are saying, before you talk to us.
No commitment. We use your details only to send your snapshot and follow up. Privacy policy
Fix the Analysis Side
Analyze comments in their original language where you can. Modern language models can tag themes and sentiment without translating first, which keeps nuance intact. When you do translate, keep the original text next to the translation so reviewers can check anything that looks odd. Our guide to sarcasm and context in feedback explains what gets lost.
Benchmark within each market
Compare each country against its own history before comparing countries with each other. Trends inside one market are more reliable than raw gaps between markets. Our post on industry VoC terminology covers vocabulary differences.
Word for word versus for meaning
Word for word
- Questions lose meaning when idioms don't translate
- Machine translation strips sarcasm and tone
- Raw scores compared across countries
Contextual (the approach the article favours)
- Translate each question for meaning
- Native speaker reviews each version
- Analyze comments in the original language
- Benchmark within each market
Frequently asked questions
Why does multi-language VoC fail?
Word-for-word translation changes the meaning of questions and strips nuance from answers.
Should you translate feedback before analyzing it?
Where possible, analyze comments in their original language to keep tone and context.
Can you compare NPS across countries?
Carefully. Scoring habits differ by culture, so compare each market against its own history first.
Who should review translated surveys?
A native speaker who understands both the language and your product.
What is a shared theme library?
It's one set of themes that feedback in every language maps to, so issues add up across markets.
How common are multilingual surveys?
About one in five surveys included at least one translation in 2024.
