What a translation round costs
Defaults are 2025-typical rates for English-source translation, and every tier card states its per-word rate. Start here, then reprice the exported CSV against vendor quotes.
All-in total
$19,578
4 locales · 20,000 words
Translation program
$10,378
Translation + LQA + PM overhead
Engineering (one-time)
$9,200
i18n framework · Webfonts · Per-locale integration
Where the money goes
- Translation$7,520(38%)
- LQA$1,504(8%)
- PM overhead$1,354(7%)
- Engineering$9,200(47%)
Parallel duration
6 days
All locales at once
Sequential duration
23 days
One translator across all
Per-locale breakdown
| Locale | Rate × | Translation | LQA | Subtotal | Days |
|---|---|---|---|---|---|
| es-ESSpanish | 1.0 | $1,600 | $320 | $1,920 | 5.7 |
| fr-FRFrench | $1,760 | $352 | $2,112 | 5.7 | |
| de-DEGerman | $1,920 | $384 | $2,304 | 5.7 | |
| ja-JPJapanese | $2,240 | $448 | $2,688 | 5.7 | |
| Translation subtotal | $7,520 | ||||
| LQA subtotal (1 round) | $1,504 | ||||
| PM overhead (15%) | $1,354 | ||||
| Translation program | $10,378 | ||||
Vendors usually set a per-locale minimum charge of $50–100. This tool applies $50 to short word counts.
1 · Source word count
Planning figures: small app ~5,000 · web product ~50,000 · enterprise app ~200,000+
2 · Translation tier
Adds throughput but increases coordination overhead.
2025 planning figures: 15-20% for in-house, 25-30% for a full-service vendor. Replace with the real PM load if known.
4 · Target locales (4)
Or load a preset
See the reach this set buys in the launch planner The planner treats these locales as already supported and ranks what to add next.
5 · Translation memory
Repetitions and fuzzy matches bill below the new-word rate.
6 · Engineering setup
One-time localization-engineering work the per-word rate never covers. The lines below come from the locale set.
- i18n framework$6,000
- Webfonts (1 script)$1,200
- Per-locale integration (4)$2,000
One-time total$9,200
Before this goes to finance
- Count only the billable words. On an update that is the delta; translation memory should cover the rest at close to zero.
- Tier moves the total more than locale count. Eight locales at MT post-edit often costs less than two at premium, so drop a tier for high-volume, low-stakes content like support and docs.
- One LQA round catches most style issues. Save a second for regulated or high-stakes launches, and a third for hero copy.
- Parallel days assume every locale starts at once; sequential is the worst case. Budget closer to sequential when one shared reviewer serves all locales.
Related: Atlas launch planner · prefills this estimator from a market shortlist · String audit · check the files the round produces