Translation Cost Calculator
Calculate translation cost from word count and per-word rate. The standard pricing model for professional translation services — used by freelancers, agencies, and clients to quote, budget, and verify invoices.
Last updated: May 2026
Compare with similar
About this calculator
The formula is total cost = wordCount × ratePerWord, where wordCount is the number of words in the source text and ratePerWord is the translator's or agency's quoted rate in your currency (USD, EUR, GBP). The result is the total fee before any minimum-charge or surcharge adjustments. Variables: wordCount from the source document's word count (most word processors and CAT tools count words automatically); ratePerWord from the translator's quote or agency rate card. Typical 2024 rates: agencies in Western markets charge $0.10–$0.25 per source word for common language pairs (English↔Spanish/French/German); rare language pairs or specialised domains (legal, medical, patent) command $0.20–$0.50+. Freelance rates are often 30–50% lower than agency rates because they exclude the agency's project management overhead. Some translators charge per target word (count after translation) rather than per source word — important to clarify upfront because target-word counts can differ 10–30% from source for languages with different morphology (English to German typically gains words; English to Chinese typically loses character-equivalents). Edge cases: rates are sometimes per 1,000 words rather than per single word — read the quote carefully. Many translators have minimum charges ($25–$50) that apply to small jobs below ~500 words. Specialised content (technical manuals, marketing copy, legal documents) often has surcharges of 25–50% on top of base rates. Rush jobs (24-hour or weekend turnaround) typically add 25–100%. The calculator handles only the base rate × words; add surcharges and minimums manually.
How to use
Example 1 — Standard business translation. A 5,000-word product manual to be translated from English to Spanish at $0.12/word (standard agency rate). Cost = 5,000 × 0.12 = $600. ✓ This is the base estimate; add 20% if the content includes technical terminology (electrical, medical, legal), and a minimum project fee (typically $50–$100) may apply if your agency has one. Example 2 — Freelance translation, rush. 12,000-word marketing brochure at €0.08/word (freelance rate, English to French), with a 50% rush surcharge for 48-hour turnaround. Base cost = 12,000 × 0.08 = €960. Rush surcharge = €960 × 0.50 = €480. Total = €1,440. ✓ For marketing content, translators typically allocate 10–20% more time for tone-matching and localisation than for technical translation — verify the quoted rate includes localisation if your target market needs cultural adaptation, not just literal translation.
Frequently asked questions
What is a typical per-word translation rate?
Rates vary widely by language pair, domain, agency vs freelancer, and quality tier. For common European language pairs (English↔Spanish/French/German/Italian) in 2024: budget agencies $0.05–$0.08/word, mid-market agencies $0.10–$0.15, premium agencies $0.18–$0.25, specialist legal/medical/patent $0.20–$0.50+. Freelance rates are typically 30–50% lower than agency for the same language pair and tier — though the agency includes project management, editing, and quality control that the freelancer doesn't. Asian language pairs (English↔Chinese/Japanese/Korean) typically cost 20–40% more than European pairs due to smaller translator pools. Less-common language pairs (Icelandic, Maltese, indigenous languages) can be 2–5× standard rates. Per-1,000-word and per-character rates are alternative pricing in some markets — convert before comparing. The single most important variable is the translator's expertise level; cheap rates often produce machine-translated output with light post-editing, while professional rates fund deep linguistic and cultural adaptation.
Should I pay per source word or per target word?
Per source word is generally clearer because the count is fixed before work begins — you know the total cost when you sign the quote. Per target word makes the cost dependent on the translator's word choices and creates incentives for verbose output. Most professional agencies and standard contracts use per-source-word pricing for this reason. Per-target-word pricing is still common in some markets (especially when the target language reliably uses more words than the source, like English→German with ~10–20% word inflation), and the rate is usually set lower to compensate. Whichever convention applies, get it in writing in the quote. The most surprising bills come from contracts that don't specify, with translators silently using per-target-word counting that pushes the bill 15–25% above what the client expected. CAT-tool counts (Trados, MemoQ) usually count source words from segmented files automatically.
Does machine translation change the cost equation?
Significantly. Machine translation (Google Translate, DeepL, ChatGPT) plus light human post-editing ('MTPE' or machine-translation post-editing) typically costs 50–70% of full human translation per word. For content where errors are tolerable (internal documents, exploratory content, low-visibility marketing), MTPE is increasingly the default — quality has improved dramatically since neural MT became standard in 2017. For high-stakes content (published books, legal contracts, medical labels, brand messaging), full human translation remains the standard because MTPE still produces errors in tone, idiom, cultural reference, and nuanced terminology. Agencies increasingly offer tiered pricing: machine-only ($0.02–$0.04/word, suitable for internal scanning), MTPE ($0.05–$0.10), full human ($0.10–$0.25), and premium specialist ($0.20–$0.50). For purely cost-driven decisions, MTPE plus careful editing is often the right choice; for quality-critical content, full human is still worth the premium.
What are the most common mistakes people make budgeting translation costs?
The first is using only the per-word rate without checking for minimum charges, which can double the effective cost on small jobs (a 200-word email at $0.15/word might be quoted at a $100 minimum, an effective $0.50/word). The second is forgetting surcharges for specialised content, rush turnaround, and certified/notarised translation; these can add 50–100% to the base rate. The third is comparing per-source-word rates with per-target-word rates from different agencies without converting. The fourth is treating translation and localisation as the same thing; localisation (cultural adaptation, idiom replacement, image and example swaps) typically costs 30–50% more than literal translation. The fifth is using the cheapest available rate for high-visibility content; the cost difference between a $0.05/word machine-translated landing page and a $0.20/word native-localised one is small relative to the marketing impact. The sixth is paying word-by-word for repetitive content (manuals with boilerplate, software strings); CAT tools and translation memory systems typically discount repeated segments by 70–100%, dramatically reducing total cost on long technical content.
When should I not use this calculator?
Skip it for translation memory or CAT-tool-discounted projects where repeated segments cost less than first-time content; use the agency's segment-discount quote instead. Avoid it for hourly-rate translation (some interpreters and specialised translators charge by time, not by word). It is the wrong tool for non-text translation: video subtitling (charged per minute of video plus translation), voice-over (per minute plus rate), and document layout / DTP work all have separate pricing models. Do not use it for certified or sworn translation, which has fixed per-document fees regardless of word count in many jurisdictions. Skip it for interpretation (live speech translation), which is charged per hour or per day. And for any large or recurring translation programme, get a written quote with discounted volume rates, translation-memory leverage, and clear scope rather than computing from a base per-word number.