Can I delegate translation to AI?
Should I still pay for translation? Machine translation has improved markedly in recent years — but the gap between “being understood” and “selling” has not closed. 🌐
The translation decision is made not by language but by the job the text does: does it carry information, build trust, or create an obligation?
Short answer: machine is sufficient for information texts, a human is required for sales and legal texts. Everything in between runs on machine plus editing. ⚖️
Three text types
BU BÖLÜMÜN ÖZETİ
- Information text
- Persuasion text
- Obligation text
This is where the decision becomes clear. 📚
Information text
Manuals, technical explanations, internal documents, frequently asked questions. Machine translation can be used directly. 📘
Persuasion text
Product descriptions, landing pages, proposals, ad copy. The machine translates but does not persuade; editing is essential. 🎯
Related reading from the archive: AI content production · AI and search visibility.
Obligation text
Contracts, warranties, regulation, health and finance information. Here translation is a matter of responsibility. ⚖️
Where does the machine struggle?
BU BÖLÜMÜN ÖZETİ
- Sector terms
- Brand language
- Cultural equivalents
- Localisation
At four points, again and again. ⚠️
Sector terms
The same word has a different equivalent in your sector. Without a glossary the translation drifts to general language. 📖
Brand language
If you say “consultant”, the tool may say “specialist”. Inconsistency erodes the brand. 🏷️
Cultural equivalents
Literally translated idioms and jokes land oddly in the target language. This part needs rewriting. 🌍
Localisation
Dates, currency, measurements, addresses and contact conventions. These are not translation but adaptation. 🗺️
How is an efficient flow built?
BU BÖLÜMÜN ÖZETİ
- Step 1: the glossary
- Step 2: machine translation
- Step 3: editing in the target language
- Step 4: localisation check
Four steps, the same for every text. ⚙️
Step 1: the glossary
Thirty or forty terms is enough: your equivalents. This file is attached to every translation. 📋
Step 2: machine translation
The raw translation is taken. In this state it is not published; it is material. 🔤
Step 3: editing in the target language
Someone who knows the language reads and edits. In persuasion texts this step is not skipped. ✍️
Step 4: localisation check
Dates, currency, measurements, contact details. Small but trust-breaking details. 🔍
How much is gained?
BU BÖLÜMÜN ÖZETİ
- In information texts
- In persuasion texts
- In obligation texts
Measurable; not exaggerated. 📊
In information texts
The gain is large: documents that were never translated before now can be. 📈
In persuasion texts
The gain is moderate: the first draft speeds up, editing is still required. ⚖️
In obligation texts
The gain is limited: used for drafts and comparison, with approval still resting on an expert. 🛑
Building a multilingual site
Translation alone is not enough. 🌍
Keyword adaptation
Headlines are not translated but chosen afresh for the target language. That is strategy, not translation. 🎯
The technical side
Without correct language tags and cross-links, the translation stays invisible. ⚙️
What should I do today?
BU BÖLÜMÜN ÖZETİ
- Step 1: split the texts into three
- Step 2: write the glossary
- Step 3: name an editor
- If you want help
Three steps, one day. 🪜
Step 1: split the texts into three
Information, persuasion, obligation. The decision comes out of that list. 📚
Step 2: write the glossary
Thirty terms, two columns. Written once, used always. 📋
Step 3: name an editor
Who reads the persuasion texts? Without a name, the step gets skipped. 👤
If you want help
Let us set up your multilingual content flow: use the consult your expert form. For your current usage see the business AI usage audit; the whole sits on the AI consultancy page. 🎯
📝 Notes From the Field
A firm published its English site entirely with machine translation. The texts were understandable but no enquiries arrived. Only four sales pages were rewritten in the target language; everything else stayed as it was. The enquiries started coming from those four pages.
📖 Short Glossary
Glossary: the two-column list of your preferred equivalents. Localisation: adapting dates, currency, measures and contact conventions to the target country. Persuasion text: text intended to influence a purchase decision. Raw translation: unedited machine output.
⚡ Quick Summary
The translation decision follows the job the text does: information, persuasion, obligation. 🌐 Machine suffices for information, needs editing for persuasion, and requires expert approval for obligation. The flow starts with a glossary and ends with a localisation check.
🎯 Next Step
Let us set up your multilingual content flow: use the consult your expert form. The verification side sits in the verification guide; for your current usage see the usage audit.
Frequently Asked Questions
Sık Sorulan Sorular
The persuasion text: the translation looks correct but has no selling force. The error is invisible on the page and visible in the results. 📉
In the combined translation-plus-editing time. Looking only at translation time misleads. ⏱️
It falls short on the search side: the phrases people actually search in the target language are different. 🔎
Selling pages by human hand; blog and support content on machine plus editing. 🪜
Quality is markedly higher between widely resourced languages; term and tone errors rise in less resourced ones. Whatever your target language, the editing rule for persuasion texts does not change.
Change their role rather than dropping them: the translator now edits and adapts instead of writing from scratch. That lowers cost while protecting quality.
It helps visitors but falls short on search and damages brand language. Publishing separate, edited content for sales pages produces better results.
Source: Google Search Central — localised versions and language tags