The Ad Localisation Process: A Practical Guide for Global Campaigns

You have an ad that works. It tested well, the production quality is high, the creative team is proud of it. Now it needs to run in a dozen markets, and in half of them the talent does not look right, the language is wrong, the pricing on the end card needs changing, and the regulatory disclaimer is different.

The ad localisation process is how a single piece of advertising gets adapted across markets, languages, audiences, and regulatory environments. It sounds straightforward in a brief and becomes exponentially complex in execution, because every market adds variables and every variable interacts with every other one.

This guide walks through how it works, where AI has changed the economics, and the practical lessons that make the difference between a smooth rollout and a painful one.

What localisation actually involves

Localisation is not one task. It is a set of overlapping production activities, and any combination might be required.

Language. Translating voiceover, text, supers, and end cards. At its simplest, subtitle insertion. At its most involved, a full rewrite of the script for cultural tone and idiom. A direct translation of English advertising copy into Portuguese or Japanese almost never sounds right. The script needs to be localised, not just translated.

Talent and representation. Changing the people in the ad to reflect the local audience. Some markets are particularly sensitive to whether the creative genuinely reflects the local ethnicity and culture. Getting this right is not just a technical exercise. It requires understanding what "representation" means in each specific market. AI face-swap localisation has made this dramatically faster and cheaper.

Visual and cultural adaptation. Product packaging with local language, pricing in local currency, culturally specific settings or props, colour associations that carry different meanings by market.

Format and duration. A 30-second 16:9 broadcast spot might need to become 20 seconds in one market, 9:16 vertical for another, 1:1 square for a third.

Music and audio. Licensing restrictions across territories, cultural resonance differences, and different broadcast loudness standards.

Regulatory compliance. Different markets, different rules. Pharma, financial services, food, and alcohol are particularly affected, with disclaimers, claims, and visual requirements that vary by territory.

The traditional process

Hero ad produced and signed off. Market-by-market brief specifying changes. Script translated. Voiceover recorded per market. Talent reshot if needed. Text and graphics replaced. Every variant through technical QC, linguistic review, and compliance check. Exported to spec and trafficked.

In a paragraph, that sounds manageable. Across 14 markets with different regulatory frameworks, three languages nobody on the team speaks, and a master that keeps being revised, it becomes a project management challenge that consumes weeks and budgets.

Where AI has actually changed things

Face-swap for talent localisation. Instead of reshooting with local talent in every market, AI replaces faces frame by frame while preserving the original performance, direction, and cinematography. A typical multi-market face-swap project runs across 3-12 markets and delivers in under five working days. The cost per market variant is a fraction of a reshoot.

AI voice synthesis and lip-sync. Translated voiceovers synthesised from the original performance, with lip movements adjusted to match the new language. The quality is now good enough for many applications, though premium brands often prefer human VO with AI lip-sync rather than fully synthetic voice.

Automated format adaptation. Resizing and reframing across aspect ratios (16:9, 9:16, 1:1, 4:5) with intelligent cropping that keeps the subject in frame. This used to require a manual re-edit for every format. AI handles the bulk of it, with human review for edge cases.

Text and graphics localisation at speed. Automated replacement of on-screen text, end cards, pricing, and legal copy. Templates make this faster and more consistent across variants.

The cost picture

Traditional full localisation (language, talent, graphics, QC) per market: £15,000-40,000. AI-assisted localisation per market: £3,000-10,000, depending on scope. The range reflects the enormous variability in what "localisation" means. Language-only text swap for three markets is a different proposition from full face-swap, VO, graphics, and compliance across twelve.

The meaningful comparison is not per-market cost but total campaign cost for the same outcome. If the traditional approach makes five markets economically viable, AI might make twelve viable for the same budget. The brand's campaign reach expands without the budget expanding.

QC: where everything either holds together or falls apart

Quality control in localisation is not a final check. It is a continuous process that runs alongside production. Every variant needs technical QC (resolution, codec, audio levels, safe areas), linguistic QC (translation accuracy, tone, cultural appropriateness), visual QC (face-swap consistency, graphics alignment, colour matching), and compliance QC (regulatory text, disclaimer accuracy, market-specific rules).

The most expensive QC failures are the ones that reach broadcast. A mistranslated end card, a face-swap artefact in a single frame, a missing disclaimer in a regulated category. These are career-affecting problems for the people who signed off. A robust QC process is not optional.

Practical lessons from doing this repeatedly

Design for localisation from the start. The most expensive problems come from master assets produced without localisation in mind: baked-in text, culturally specific dialogue, visual elements that do not work in certain markets. A five-minute conversation at the briefing stage prevents a five-week problem in post.

Keep a single master. When the master changes, changes must flow to all variants. If each market version is an independent file, a late-stage master revision becomes 14 separate re-edits.

Centralise the workflow. One partner managing translation, VO, face-swap, graphics, QC, and delivery. Each vendor handoff is a potential failure point.

Build a localisation kit. For every master: ProRes export, clean plates, source-language script, list of every text element needing localisation, specs per market, market-by-market scope brief. This takes an afternoon to prepare. Not preparing it costs weeks.

Questions that come up

How long? Language-only for a few markets: 1-2 weeks. Full localisation across 10+ markets: 2-4 weeks. AI workflows are 40-60% faster.

Can you localise something not built for it? Yes, though quality and cost depend on the master. Clean project files are cheaper to work with than flattened assets. We will assess and give you a straight answer.

Do I need new talent contracts for face-swap? Yes. The original talent's agreement must permit AI modification. This needs to be in writing.

Can planning start before the master is finished? Yes, and it should. Translation, briefing, and format planning can run in parallel with master production. Face-swap and VO sync need a near-final master. Starting planning early compresses the overall timeline meaningfully.

Localisation is where global ambition meets local complexity. The variables multiply with every market. AI has made the production mechanics dramatically faster and cheaper, but the strategic and cultural work, the translation craft, the sensitivity to local representation, and the lip-sync precision still require human skill and experience.

If you have a campaign that needs to work across borders, we can help you work out the fastest route from one master to many markets.