Localization, as a corporate function, is in a strange transitional moment. The translate-first vendor model that has dominated since the late 1990s — large globalization vendors managing translation memory across dozens of languages and routing work to networks of contracted translators — is being out-competed in three directions at once. By in-market editorial teams who treat localization as transcreation. By AI-augmented workflows that handle volume work at a fraction of the cost. And by personalized content systems that obviate the question of localization entirely by generating contextually appropriate content at request time.
The vendor model isn't going to disappear, but its place in the brand stack is shrinking quickly. The brands that read most native in their international markets are no longer the ones with the largest translation budgets. They are the ones with the best in-market editorial talent, augmented by infrastructure that handles volume.
Why translate-first stopped working.
Translate-first works when the source content is generic enough that a faithful translation produces appropriate destination content. The problem is that most modern brand content is highly contextual — it references cultural moments, uses register and tone deliberately, and makes choices that are meaningful in the source culture and meaningless or wrong in the destination culture.
A faithful translation of a tongue-in-cheek US email subject line into Japanese produces an email subject line that is grammatically Japanese and culturally invisible. The Japanese reader will not be offended. They will simply not read it. Multiplied across an entire content program, this represents a slow leak of brand meaning that compounds into customer indifference.
What transcreation actually is.
Transcreation, properly practiced, is a different discipline from translation. The transcreator is given the source brief, the source content, and the editorial intent — and produces destination-language content that delivers the editorial intent in the destination market, with whatever creative reinvention is required. It is closer to copywriting than to translation.
Transcreation is more expensive per unit than translation. It is also more effective per unit by enough margin that the unit economics often favor it for high-leverage content surfaces — hero campaigns, brand films, product launch content, organic social. For high-volume low-leverage content — product specifications, support content, shipping confirmations — pure transcreation is overkill.
AI-augmented workflows are real now.
Two years ago, AI translation was 'almost good enough' for most consumer content and required heavy human editing to be brand-safe. Today, AI translation, when prompted with brand voice context and editorial values, produces output that needs only light editing for many content types. The differential is large enough that the operational model around localization should be redesigned.
What works in our experience is a tiered model. AI handles first-pass production for operational content. In-market editors review for brand fit and cultural appropriateness. Transcreation specialists own the high-leverage content end-to-end. The operational team manages the workflow rather than producing the work. The total cost of localization drops while quality at the high-leverage end actually improves.
Notably, the brands moving fastest in this direction are the ones with strong in-market editorial talent — because the AI requires editorial supervision to be brand-safe, and the in-market editor is the irreplaceable component. The brands without in-market editorial capability are at risk of using AI translation to scale content that is uniformly bad, faster than ever.
Personalization changes the question.
The third pressure on translate-first localization is the rise of contextual content generation at request time. Email content tailored to the recipient's behavior. Product detail pages assembled from modular content components. Onboarding flows that adapt to detected user signals.
When content is generated dynamically, the question of localization shifts. You are not localizing assets; you are localizing rules and components. The localization function evolves into a content engineering function — building the component library, the rules, and the brand-safety guardrails that allow personalized content to be produced reliably across markets.
This is closer to product engineering than to traditional localization vendor management. Brands building this capability tend to organize it inside marketing engineering or growth ops rather than inside the localization function. The traditional localization function is, in many of these brands, becoming a center of editorial excellence rather than a production function.
"Localization stopped being a translation problem and started being a content engineering problem. The brands that noticed are pulling away."
What to do this year.
If you are running localization for a cross-border brand, four moves are worth making in the next twelve months.
- Stratify your content by leverage. Decide which surfaces deserve transcreation, which deserve AI-augmented translation with editorial review, and which can run on pure machine translation.
- Invest in in-market editorial talent. The single highest-leverage role you can hire is a senior editor in your most important international market who owns brand-safety across all content tiers.
- Build the AI workflow with brand-voice context, not as a bolt-on. Generic AI translation produces generic content. Brand-aware AI translation, properly prompted with editorial values and reference content, produces content that needs minimal editing.
- Start treating localization as a content engineering capability. The brands organizing this work as engineering rather than vendor management are pulling ahead measurably.
The translate-first vendor model isn't going to be wrong. It is going to be insufficient. The brands building the next localization capability — editorial talent, AI augmentation, content engineering — are quietly acquiring an advantage that will be hard to replicate once it compounds.
Deebo is an international expansion and cross-border ecommerce agency, headquartered in Tokyo, working with foreign brands across Japan, APAC, and 40+ markets worldwide.