Language service costs vary by language pair, broadly mapping to the relative GDP of the market, with exceptions for rarer languages where demand outstrips supply. Word counts can be reduced through weighted counting and translation memory, though this only works within a language, not across languages, and depends on how repetitive the content is.
Output quality is set out as a mix of objective measures, spelling, grammar, and channel compliance, and subjective ones, tone of voice and cultural alignment, with customer engagement data treated as the most reliable indicator of whether content is working. Six content creation methodologies are compared, from machine translation with human post-editing through to copywriting, each carrying different cost and quality trade-offs.
Word rates work for simple, repetitive content, but fall down for anything requiring keyword research, cultural nuance or ongoing optimisation. A performance-driven alternative feeds online KPI data into production and prices by time or output rather than by word. Machine translation lowers cost for high-volume content, but still depends on human post-editing for most business use cases, and that effort is often unclear in supplier pricing.
Procurement teams are given questions to ask suppliers throughout, closing with a call for pricing that reflects the actual value added at each stage of production.
Download Whitepaper