How to
How to translate merch program copy
A prose guide — what goes into the glossary, what never gets translated, how each language gets typographically tested — so a trilingual program in Mexico City, Monterrey or Guadalajara reaches the recipient as the same brand in three languages, not as three different translations of the same text.
Why most trilingual merch programs sound like three different brands
The prose of a merch program's translation does not fail because of the language; it fails because of the vacuum of instructions. The brief the translator receives says "translate to English" or "localize to Russian" without giving them the brand glossary, the list of terms that must not be touched, the audience the message is aimed at and the registers the brand has already prohibited. The translator, who is serious, does their job and returns a technically correct translation that wrecks the brand voice by the second sentence. The brand reaches the recipient as three distinct brands that casually speak three languages.
In Mexico the problem intersects with three particular frictions. The first is the gap between Mexican Spanish and European or general Latin American Spanish: a program that is commissioned from Madrid or Buenos Aires and that lands in Mexico City runs the risk of sounding foreign if it is not reviewed by a local native. The third is the negotiation of English: for a trilingual Spanish-English-Russian program, English works as a pivot language in the glossary, but the final version of each market is decided by a native of the market, not by machine translation or by the central team. The third is the brand register in Russian, which demands decisions on transliteration of the proper name, on the order of given name and surname, and on the use of Cyrillic or Latin script depending on the channel.
This guide is for the brand or marketing team that has to translate a program before the next cycle. It covers the prose of the seven steps to produce a translation that holds the brand voice across the three languages, and the antipatterns that turn the translation into a post-print negotiation. It does not discuss the CAT tool or the decision to machine-translate; only the prose of the process that protects the brand voice.
The intended scale is a program of between five and twenty copy pieces per language — slogans, claims, packaging microcopy, delivery emails, user manuals when applicable — distributed across one to three target markets. The structure holds for smaller and larger programs; what changes is the size of the glossary and the review frequency.
The translation process in 7 steps
- Brand voice brief per language. One page per language with the voice description (warm, technical, luxurious, irreverent), examples of phrases that are allowed, examples of phrases that are not allowed, and the brand persona in that market. The brief is not optional; without it the translator decides on their own.
- Closed glossary of brand terms. A spreadsheet with each brand-specific term — product names, registered slogans, invented words, internal toponyms — and the translation decision for each one: "kept in original", "transliterated", "translated with footnote gloss", "localized with local term". The glossary is signed before starting the translation.
- Translation by a native of the target market. Each language is translated by a native of the target market, not by a native of the language working in the target market. A native of Moscow living in Mexico City translates better to the Russian of Moscow than a native of Saint Petersburg living in Buenos Aires. The process's prose records who translated and from where.
- Typographic test per SKU. The translated copy is mounted on the SKU mockup (t-shirt, mug, notebook, packaging) with the final typography and body size. The typographic test detects extensions and contractions that change the composition: Spanish takes between 18% and 25% more characters than English; Cyrillic Russian is often shorter than English in the same structure.
- QA review per market. A native of the target market reviews the final version against the voice brief, the glossary and the typographic test. The review is not optional; without it the final version may pass as correct without being so. The QA's prose signs each piece.
- Final approval by the brand owner. The program's brand owner reviews the three versions in a single comparative view, with optional audio when the copy goes to a video or podcast. The approval closes the copy phase and opens the final art phase.
- Reference archive for the next cycle. The glossary, the voice brief, the translation decisions and the QA reviews are kept in a folder accessible to the next cycle. The archive sustains brand consistency and cuts the time of the next translation.
Prose antipatterns that break the brand voice
The first antipattern is translating to the language without translating to the market. "English" is not a market; it is a language with natives in Toronto, Mumbai, Manila and Lagos who expect different things. The brief's prose distinguishes between language and target market and requires a native of the market, not of the language.
The second antipattern is the implicit glossary. If the brand terms are not written down, the translator improvises them. The glossary's prose is closed and signed before starting the translation; any exception is approved in writing.
The third antipattern is machine translation without review. Machine translation is a tool, not a translator; without a native who reviews, it returns a version that sounds like a machine. The process's prose prohibits the delivery of automatic copy without a native's signature.
The fourth antipattern is ignoring typographic extension. A claim that fits in two lines in Spanish can blow up the composition in English or Russian. The process's prose demands the typographic test before the final approval, not after.
The fifth antipattern is monolingual approval. If the brand owner approves only the Spanish version and trusts the rest, the English or Russian brand reaches the recipient without a voice filter. The process's prose demands a final approval that covers the three versions in a comparative view.
The sixth antipattern is the glossary that no one re-reads. A glossary archived in a folder that the next cycle does not open loses consistency. The process's prose assigns someone to re-read the glossary and propose updates to the brand owner.
The seventh antipattern is the lack of QA budget. If the program's budget does not include the QA step by a native, the program is delivered without it. The program's budget prose reserves an explicit line for QA per language.
Tools, templates and references
Voice brief per language. One page per language with voice description, allowed and prohibited examples, and brand persona. The brief is signed before starting the translation and is delivered to the translator together with the glossary.
Closed glossary. A spreadsheet with source-language term, decision (kept, transliterated, translated, localized), target-language term, and observations. The glossary is signed by the brand owner and by the translator.
Typographic mockup per SKU. An editable file with the translated copy mounted on each SKU, with the final typography and body size. The mockup is reviewed before the final approval and is archived alongside the final art.
QA checklist per market. One page per language with ten to fifteen criteria reviewed by the market native: voice, glossary, typographic extension, register, locales, diacritic marks, numbers, dates, units and postal address.
Comparative approval view. A table with the three copy versions side by side, in separate columns, with the source-language versions on top as reference. This is the view the brand owner reviews to approve.
Cycle reference archive. A folder with the voice brief, the glossary, the translation decisions, the QA reviews and the approvals. The archive lives within reach of the next cycle and of the central brand team.
Useful references. RAE Dictionary of the Spanish Language for Spanish register; DPD (Pan-Spanish Dictionary of Doubts) for regional variants; Fundéu BBVA's style guide when applicable; Merriam-Webster for US English register; Oxford English Dictionary for British register; Oxford Russian Dictionary or Gramota for standard Russian; ISO 8601 for dates and times; RFC 5646 for locale tags (es-MX, en-US, ru-RU).
Suggested schedule. Brief and glossary in week 1; translation in weeks 2 and 3; typographic test in weeks 3 and 4; QA by native in week 4; final approval in week 5; reference archive at the close of week 5. Five weeks to hold the brand voice across three languages.