Positioning is Marketing

Localisation Isn’t Translation: How Cultural Intelligence Builds the Right Market Positioning

AI translates your words. Language experts translate your meaning.

Image generated using Gemini

There is a phrase I use in every GTM briefing I run: “Localisation isn’t translation.”

Most people nod. They have heard it before. They believe it in the abstract.

Then they ship a product into a new market, watch the numbers come in, and discover that believing something in the abstract is not the same as understanding it in practice.

I have built marketing and positioning strategies across four continents. A live news platform in the Indian Ocean region. A B2B product spanning twelve linguistic markets across India. A luxury brand entering Japan. In each case, I arrived with a framework. In each case, the market pushed back and forced me to build a better one.

This piece is about that evolution and the methodology that came out of them.

Where it started: the AI pipeline that taught me what AI cannot do

Image generated using Gemini

In 2021, I was building the content and marketing operations for a traditional media organization going digital. Their first platform in a dual-language B2C market, targeting over one million customers across French-speaking regions. And live news operations meant speed was non-negotiable. A publication cannot wait 48 hours for human translation at scale.

The solution was clear: an AI-assisted pipeline. The efficiency gain was immediate. 20% faster turnaround, significantly reduced bottlenecks, and an editorial team that could focus on creation rather than conversion. On paper, the pipeline worked exactly as I designed it to.

Then a reader wrote in.

A phrase in one of the articles translated in French had caught their attention. Linguistically, there was nothing wrong with it. But in the context of the specific political history of Mauritius, it was a charged expression. One that carries connotations in the Indian Ocean region that it simply does not carry in metropolitan France.

The model had no framework for that distinction. It was not a hallucination. It was not a mistranslation. It was a failure of contextual awareness that no amount of linguistic training data could have prevented, because the issue was historical and political, not linguistic. The piece required a formal correction.

That experience built the foundation of everything that followed. I restructured the pipeline: AI for speed and volume on standard content, local editorial review with embedded cultural knowledge for anything touching politics, policy, social identity, or historical reference.

The tool knows the language. It does not know the audience. There is a category of intelligence, the kind that comes from living within a cultural moment, that no model trained on text can fully replicate. That is where the human layer earns its place.

This was lesson one. But it took a second project to show me how far it actually went.

Where it deepened: churn, onboarding, and the expressions that hold users

Image generated using Gemini

The second project was a large-scale B2B market expansion in India. Twelve distinct linguistic regions, over five million vendors, an aggressive timeline.

After the initial GTM, the data showed a pattern I had not anticipated: a higher churn rate than the market benchmarks suggested we should be seeing. Users were activating but not staying. The marketing was reaching them. Something in the experience was failing to hold them.

The diagnosis led me somewhere most positioning briefs don’t go: onboarding flows, UX copy, and knowledge base articles. Not the campaign content. The product itself.

This is where most platforms get it wrong, and get it wrong expensively. When a company enters a new market, the instinct is to translate the interface and move on. The homepage gets localised. The campaign copy gets localised. The onboarding tooltip that a new user encounters at their most uncertain moment, the error message they receive when something doesn’t work, those get run through an AI pipeline and shipped.

This is where you lose users. Not at the top of the funnel. Inside the product, at the moments that matter most.

For this project, we didn’t just translate the microcopy, the UX copy, and the knowledge base articles, but brought in language experts. Not translators, but people embedded in the professional culture of each region.

The question was not “is this accurate?” but “does this expression mean the same thing here, in this context, to this user?”

That distinction sounds subtle. It isn’t. A phrase that reads as encouraging in standard Hindi can read as condescending in a state where formal register carries cultural weight. An onboarding instruction that feels intuitive in one region can feel ambiguous in another because the implied relationship between the platform and the user is different. These are not language errors. They are experience errors which compound over time leading to unprecedented leakage and higher churn rates.

The restructured approach worked. Language experts who understood not just the words but the expressions, the specific turns of phrase that carry meaning in a particular community, rebuilt the onboarding and knowledge base from the inside. The result was not just a more accurate product. It was a product that felt like it had been built for those users, not adapted for them.

There is a meaningful difference between those two things. Users feel it immediately, even if they cannot name it.

Where it arrived: positioning as a cultural act

Image generated using Gemini

In 2025, during my MBA at ESCP Business School in Paris, I worked on the market entry strategy for a French luxury brand planning to expand into Japan. The financial modelling was sound. The positioning was clearly articulated. The product was genuinely strong for the target segment.

In user testing with Japanese stakeholders, the feedback was consistent across every participant: “I understand what this says. It doesn’t feel like it is for me.”

The language was not the problem. The language was technically flawless.

What the testing revealed was something the standard framework doesn’t account for. In Japan, nuance isn’t just about what you say. It is about the expression you choose, the context in which you place an idea, and what that placement communicates about your relationship to the reader. Where information sits on a page, how directly a value proposition is stated, whether heritage is invoked as aspiration or a mere placeholder. All of it carries meaning that has nothing to do with translation.

The emphasis on founder narrative and heritage, powerful in French luxury positioning, landed as institutional and remote rather than aspirational and personal. The imagery felt imported. The positioning had been converted accurately from one language to another, but the underlying cultural logic had not been adapted.

We hadn’t localised the message. We had translated a European positioning framework into Japanese words.

The work that followed was not about changing the copy. It was about rebuilding the positioning from the perspective of a Japanese consumer’s relationship to luxury, status, and personal expression, and then finding language that embodied that framework rather than the original one.

This is what cultural intelligence looks like at the positioning level. It is not a finishing step applied to a completed strategy. It is a foundational input that shapes the strategy itself. The question is not “how do we translate this message?”, but “how would someone embedded in this culture, in this market, would have written this message in the first place?”

What this means for every platform entering a new market

Image generated using Gemini

AI translation tools are genuinely powerful. They are indispensable for speed, volume, and linguistic accuracy at scale. I still use them. The pipeline I built in 2021 still runs on the same hybrid principle.

But there is a category of work that AI cannot do, and the gap is not closing as fast as most product teams assume.

The gap is not in vocabulary or grammar. It is in the contextual awareness that comes from being embedded in a community: knowing which expressions carry professional weight in a specific region, understanding how a UX instruction lands differently depending on the implied relationship between platform and user, recognising when the placement of an idea on a page communicates something the words themselves do not.

Most platforms entering new markets are using AI to translate their onboarding flows, their microcopy, and their knowledge base. They are shipping linguistically accurate products that feel foreign to the users they are trying to retain. The churn shows up in the data weeks later, attributed to product-market fit or activation issues, when the real cause is something much more specific and much more fixable.

Language experts, not translators, but people who are culturally embedded in the market you are entering, do three things that AI cannot.

1. They get the expressions right, not just the words.
2. They understand placement and context, the way an idea needs to sit within a user’s cultural frame to land the way it was intended.
3. And they flag before you ship, not after a user writes in.

The difference between a platform that retains users in a new market and one that doesn’t is often not the product. It is whether the product speaks to those users in a way that feels like it was built for them.

The question worth asking before the next launch

The standard localisation checklist (translate the content, adapt the imagery, localise the pricing) is not wrong. It is just heavily insufficient.

Image generated using Gemini

Cultural intelligence is not a layer you apply on top of a finished positioning. It is the input that determines whether your positioning was built for the market or merely translated into it.

Before your next international launch, ask one question about every piece of positioning, every onboarding flow, every piece of UX copy you plan to take to market:

Would someone born and raised in this market, embedded in its professional culture and its history, have written this, in this way, in this order, with these expressions, themselves?

If the answer is yes, you are localising.

If the answer is no, you are translating.

And the difference between those two things is, eventually, what your retention numbers will tell you.

What has been your experience with localisation at scale, and where did the standard approach fall short?