Help-center translation that stays in step with your articles
Your help center changes every week, and each language falls behind unless someone keeps up. We keep up for you, so customers in every language get the same answer as your source readers, in the same product terms. Our help-center translation covers knowledge-base articles, support macros, and other support content. Agents find what changed, a native linguist works on it at the level each article needs, and an automated quality gate checks every file before delivery. We work from your exports, such as HTML, XML, JSON, or XLIFF files, or inside your translation management system (TMS).
Keep every language in step with your articles
Support content works only when the answer in Spanish matches the answer in English. We translate on your update cycle, so a fix in a source article reaches every language you publish. Platforms we work inside include Phrase, XTM, Smartling, Crowdin, and Lokalise. Articles come back in your format, with tags and placeholders preserved.
Send one large batch to clear your backlog, then a steady stream of updates as your articles change.
Updates, not retranslation
Each update cycle starts with what changed. We compare each source release with the previous source version. We translate what changed and recheck existing translations affected by changes in context, approved terminology, or product behavior. Approved translations stay in your translation memory (TM). Decisions you approved stay closed. They reopen only when you ask, or when new evidence appears.
Product terms that match your UI
When an article calls a button by a different name than your product does, customers get stuck and open a ticket. Terms come from your glossary, your TM, and your published content first, and every term carries its source. The quality gate checks locked terms on every delivered file.
If we also localize your product UI, both draw on the same glossary, so the article and the screen use the same words.
Matching depth to article priority
Not every article carries the same risk. At the start, we agree priority tiers for your articles, and each tier gets a service level:
- Archived or low-traffic articles: AI + Check, if you choose it. A native linguist checks every flagged segment plus a random sample.
- Current articles and support macros: AI + Edit. A native post-editor edits every segment.
- Billing, legal-policy, safety, and account-security articles: AI + Edit + Review. A second linguist reviews every segment.
You can also keep specific articles out of AI. Those articles are translated by native linguists without machine translation or generative AI, using your translation memory and glossary.
Every delivery includes the quality-gate result and an issue log. At AI + Edit, sampled quality reports follow the schedule agreed for your program and can cover several small releases; at AI + Edit + Review, each delivery gets a full report. Pooling small updates gives the sample enough text to be useful. Glossary changes and a query log come with AI + Edit and AI + Edit + Review deliveries only.
Your team’s edits make the next batch better
Your support team will edit some translations after they go live. We want those edits, because each one is a lesson we apply before your next batch. After each delivery, we collect your edited files or the published version.
Every edit your reviewer makes is compared with what we delivered. Each confirmed lesson becomes a glossary entry, a rule or an automatic check before your next batch.
When your reviewer’s edit is better than our fix, we say so. If you find a language issue within the agreed scope, we correct it free of charge.
Questions about help-center translation
How do you get our articles?
We work from the article exports your help-center platform produces, or inside your TMS, including Phrase, XTM, Smartling, Crowdin, and Lokalise. Articles come back in your format, with tags and placeholders preserved.
Can we keep some articles out of AI?
Yes. Those articles are translated by native linguists without machine translation or generative AI, using your translation memory and glossary.
Will product names and UI terms stay consistent?
Yes. Terms come from your glossary, TM, and published content, each with its source, and the quality gate checks locked terms on every delivered file.
Do you translate support macros?
Yes. Support macros usually run at AI + Edit, where a native post-editor edits every segment.
Can we try the service before committing?
Yes. Request a quote, and if the fit is right, we run a sample on your own articles at the level we propose. If you already have AI output, a free check on up to 300 source words can come first. There is one free check per company.