Axe launches to Gen Z in the audience's own language, with applied AI and Recommend
There is a specific way brands fail with Gen Z: they learn what the cohort cares about, then describe it in the brand's own words. The topics are right, the phrasing is wrong, and the audience recognises an outsider's impression before the first sentence ends.
For Axe's launch into Croatia and Slovenia, the system treated those as two different problems. Applied AI tracked the cohort's topics — and, separately, its vocabulary: the actual phrases the audience uses. Topics decided what the communication was about. Vocabulary decided how it sounded. Seven months of launch communication, in the audience's own words the whole way.
What changed
Topics and vocabulary tracked as two separate inputs — what the cohort discusses, and how it says it
Seven months of launch communication that stayed current across two markets
Everything running on Recommend's own infrastructure, with each product linked through to the retailer
A short path from earned attention to a completed purchase at dm
Fluency is the credibility test
Topic tracking answers what an audience cares about. It cannot answer how they talk about it — and with a cohort that treats language as a membership test, that second question decides whether anything gets read.
So vocabulary ran as its own tracked input: not the brand's translation of what the audience means, but the phrases they actually use, kept current as they changed. It is a finer-grained method than topic tracking alone, and it is specifically what keeps seven months of communication from drifting into an outsider's voice.
A launch that had to stay current
A launch burst is easy — everything is fresh for three weeks. This launch ran for seven months, across two markets, into an audience whose topics and phrasing turn over constantly.
That is an infrastructure problem, not a creative one. The communication was organised around the audience and produced continuously, so month six sounded as current as week one — because the inputs it was built from were still being read, not remembered.
Our infrastructure, the retailer's checkout
The communication ran on Recommend's own infrastructure, with every product connected through to dm — so a reader moved from the content to the retailer's shelf in one step.
That structure splits the job honestly: the communication has to earn attention in its own right, and then carry the reader to where the purchase completes. It suits any brand that sells through a dominant retailer — the brand gets a home for its story without negotiating space on someone else's platform, and the retailer receives qualified traffic.
A method for any cohort with a dialect
Gen Z is the sharpest case, not the only one. Developers, gamers, athletes, professionals — every cohort with its own dialect applies the same test: do you actually speak it?
That is what applied AI makes sustainable: reading a moving audience's language as data, and producing communication in it — for as long as the launch needs, not as long as a creative team can improvise.