Match each route to a market by what it costs to say again in another language
The problem this solves. If you must address two audience languages (audience-language-is-the-market), the naive plan is to run both operations twice and go broke. The way out is that formats do not cost the same to duplicate — the spread between the cheapest and the dearest is enormous, and it does not track how well the format works.
What actually makes a format expensive to repeat in another language:
- NOT the narrator's own fluency — this was the first driver written here and it was wrong. Voice cloning ends it: a thirty-second reference sample per language buys unlimited narration in that language, in the operator's own timbre. What survives is that the script must be rewritten from the structure in each language, never translated.
- Whether the edit is timed to one specific narration track. This is the driver that actually bites. Cut the picture to a Chinese voice-over and an English script twenty percent longer knocks the whole timeline out. Cut to duration-elastic segments — holds, slow moves, static cards — and the same picture serves every language. Free to decide before the first episode, expensive after.
- Whether any text is burned into the picture. Titles, subtitles and word cards must live on a separate layer composited at export.
- How much of it is culture-borne. Structure travels; the furniture does not.
- Whether the artifact is the product itself. A feature that speaks the user's language on its own lands in any market at no extra cost.
The rule. The expensive drivers are all production-design decisions made before the first piece is shot — not facts about the operator's talent.
Where the tension is real. Duration-elastic pictures cannot be cut to a beat, and cutting to a beat is what makes short-form work. Resolve it by asset class: the long piece obeys portability; the short clips extracted from it are re-cut per language, because they are short enough that re-cutting is cheap and rhythm is where their value lives.
Share the layer above the language. The costly work is not the rendering, it is the finding — which sentence, which gap, which book, which complaint. That research is language-independent: one pipeline feeding two rendering lines. The mistake is to organize by market at the top, which duplicates the research too.
Three questions to price a route: could a person who does not speak my best language make this? does it survive being made by someone with no charisma? does the second version reuse the first version's homework? Three yeses — nearly free in every market. Three noes — a one-market bet; aim it at the market that most rewards being you.
Kin: audience-language-is-the-market · format-lag-arbitrage.