Consistency isn't uniformity: what Qonto learned about keeping one brand voice across eight markets
Why the formal "you" works in Frankfurt and backfires in Madrid, why feeding your brand guidelines to AI doesn’t always work, and what your content team does with the hours it gets back
This post is based on a conversation with Kim Reyes, Head of the Creative Factory at Qonto, recorded for the debut episode of Next Market Live: The Podcast.
The most popular myth about international expansion, according to Kim Reyes, is that you can copy and paste what you do from one market to the next. Translate the site, swap the currency symbol, keep the same tone everywhere, and trust that what charmed customers in Paris will charm them in Berlin. They’re basically neighbors anyway, right?
No, it won’t work, and Kim has spent her career proving why that fails. She’s lived in six countries, started in journalism, and now runs the Creative Factory at Qonto, the European business-banking platform that operates across France, Germany, Italy, Spain, the Netherlands, Belgium, Austria, and Portugal. Her job sounds simple enough, that is, until you try it yourself: keep Qonto sounding like Qonto in eight markets and several languages, without flattening it into something that sounds like your generic bank.
The trap most teams fall into, she says, is mistaking consistency for uniformity. They are not the same thing, and confusing them is how a brand ends up either fragmented or robotic.
The formal “you” problem
Half of Europe makes you choose a side of a grammatical line that modern English doesn’t have: the formal and informal “you.” French has vous and tu. German has Sie and du. Italian, Spanish, and Dutch all draw the same distinction. Pick wrong and you’ve either insulted someone or embarrassed yourself before finishing your first sentence.
A uniformity-minded brand would pick one rule and enforce it everywhere, in the name of staying “on brand.” Yet Qonto went the other way.
In France and Germany, where finance and business communication carry an expectation of formality, the brand speaks in the formal register. In Italy and Spain, it uses the informal. As Kim puts it, a B2B tech company addressing an Italian audience with the formal “you” would just look silly – a branding mishap a proper banking platform would want to avoid.
Instead, the brand stayed consistent precisely by changing the register. Consistency, in Kim’s framing, is brand-level code-switching. You stay recognizably yourself while adapting to the room, the same thing a person does moving between a board meeting and a dinner with friends. The version of consistency that insists on identical rules in every market is tone-deaf rather than disciplined.
Localization inside localization
Then it gets more granular, in a way that would make every localization lead wince in recognition.
When Qonto expanded into the Netherlands and Belgium, Dutch and Belgian Flemish were new languages for the team. So they did what every localization guide recommends you do: benchmarked competitors, talked to local experts, consulted native speakers on staff. The evidence pointed toward the informal register, and that matched how Qonto wanted to be positioned. Decision made, documented, implemented.
It worked on landing pages and in emails. Then the onboarding team started phoning new customers, and the informal “you” suddenly felt wrong. This was a stranger’s first live human contact with the brand, on the phone, talking about your money and your business. In that moment, the casual register that read as friendly on a webpage read as presumptuous on a call.
Even though the evidence said otherwise, Qonto went with its gut and split the decision. Informal for Dutch and Flemish marketing, formal for that first onboarding conversation, the whole thing written down with the reasoning attached. Kim calls it “localization inside localization,” and it’s a detail that, while seemingly minor, largely separates teams who localize on paper from teams who localize for how a market actually experiences them. Sometimes, the right register goes beyond the property of a language; it’s the moment that defines it.
Why you need to rewrite your guidelines before feeding it to the AI
By 2024, all of this excellence had created a very modern problem. Qonto’s content team had become so good at on-brand localization that everyone wanted “a quick localization, can you just check this?” The quick checks piled up until they were eating 30 to 40% of the local content team’s capacity. Necessary work, but not the work you hire brand specialists to do.
The obvious fix was AI, and Qonto had the tooling: an enterprise platform giving every employee access to the major models, plus the ability to build custom assistants. So the team did the intuitive thing first. They took their tone-of-voice guidelines, the ones honed for humans, and fed them straight into the model.
The results, in Kim’s words, were horrible. Terrible.
Brand guidelines written for humans are full of abstraction, general vibes, and visual shorthand. Qonto’s even used emojis as tonal signals, usually those a human being would instantly understand but a machine wouldn’t.
Humans fill the gaps with judgment. A model fills them with whatever it guesses.
Instead of trying to construct a better model, they decided to rewrite the guidelines for the machine, stripping out the abstraction until there was almost no room left for interpretation. Something that marketers who work with LLMs deeply ingrained into their workflow understand by now.
While that rewrite can be unglamorous, it’s undoubtedly the heart of the whole project, and it’s also what sets Qonto apart from companies that skip this and announce they’ve “deployed AI for content.” Your guidelines are the model’s training material. If they’re mush, the output is mush.
What the team did instead of disappearing
Once the guidelines were machine-readable, Qonto built localization assistants for each language, affectionately named Tolki, and tuned them market by market.
Naturally, while it produced clean German output, it produced garbage in French, so they had to tune it per language. They added a second assistant to grade the first one against four quality gates: grammar, regulatory compliance, tone of voice, and native fluency. A non-native speaker no longer needed to judge whether the output was any good, because the quality guard did the first pass.
The payoff was worth it and measurable. Tasks that took hours or days dropped to minutes, with time savings of 50% on some formats and 75% or more on others, at a quality Kim describes as matching or beating a junior copywriter.
Still, she recommends a human proofread at the end. The point was never to remove the human, but to stop spending the human on rote copy-pasting.
Which answers the question every content team is afraid to ask. The local teams didn’t build their replacements with AI and disappear; they got their jobs back and were able to focus their cognitive load on things that AI couldn’t do.
Before, the flood of requests had pushed them into pure execution mode, with no room to step back and think. Reclaiming those hours meant they could return to the strategic and creative work, the brainstorms, the message testing, the customer research, that they were hired for in the first place.
Kim’s advice to anxious teams is to frame AI not as the thing coming for your job but as the thing handing back the part of it you actually wanted.
A strong brand identity is easier to translate
It’s tempting to file this under “AI productivity story,” but that misses what makes Qonto’s localization journey so captivating. Every move Qonto made, from the formal-versus-informal split to the phone-call exception to the machine-readable guidelines, runs on the same principle: know your core so well that you can flex everything around it without losing yourself.
That’s the discipline beneath all of it. Brands that treat consistency as a set of frozen rules end up either sounding identical in every market, which means foreign or awkward in most of them, or letting each market drift until there’s no brand left. The teams getting it right do the harder thing: pin down what’s non-negotiable, the mission, the values, the essence that makes the brand itself, and then trust local judgment with the rest.
Uniformity is what you reach for when you don’t trust your teams or yourself. Consistency is what you get when you know exactly who you are, and let that survive the trip across the border.