Perfumer & Flavorist asked Avner Gal, co-founder and CEO of iRomaScents, how the model generates fragrance recommendations, particularly in light of the subjectivity of scent perception, the complexity of interactions among ingredients. iRomaScents
Running from July 27 through August 17, 2026, the limited-time activation invites shoppers to discover their ideal fragrance through iRomaScents' intelligent retail platform. After completing a brief AI-guided questionnaire, consumers receive tailored fragrance recommendations before sampling their matches through a touch-free smart fragrance station, offering a personalized alternative to traditional fragrance browsing.
The activation spotlights three Le Monde Gourmand fragrances, including the newly launched Banane Délice Eau de Parfum, alongside bestsellers Crème Vanille Eau de Parfum and Lait de Coco Eau de Parfum.
Leading the collection is Banane Délice, a modern gourmand built around banana milk, pineapple and orange top notes that melt into a creamy heart of Coconut Solar, Sugar Accord and heliotrope, before settling into a comforting base of tonka bean, whipped vanilla and sugared musks. Designed to evoke an endless summer escape, the fragrance blends nostalgic sweetness with a contemporary tropical twist.
As retailers increasingly look to combine personalization with measurable commercial performance, the Sephora Marbella activation serves as a real-world demonstration of how AI-powered scent matching can reshape the fragrance shopping experience while introducing consumers to new launches and best-selling scents.iRomaScents
The activation also features Crème Vanille, an elevated take on vanilla that layers creamy apricot, vanilla orchid, brown sugar and amber for a warm, signature-style scent, while Lait de Coco captures a year-round beach escape through notes of bergamot, vanilla praline and creamy coconut.
Beyond showcasing the fragrances themselves, the initiative highlights the growing role of AI in fragrance retail. According to iRomaScents, its technology can identify fragrance preferences and deliver personalized recommendations in under a minute. The company says the platform has demonstrated conversion rates of more than 90%, compared with the typical 5% to 10% conversion rates associated with conventional fragrance retail.
Running from July 27 through August 17, 2026, the limited-time activation invites shoppers to discover their ideal fragrance through iRomaScents' intelligent retail platform.iRomaScents
Perfumer & Flavorist asked Avner Gal, co-founder and CEO of iRomaScents, how the model generates fragrance recommendations, particularly in light of the subjectivity of scent perception, the complexity of interactions among ingredients.
"The system walks the user through variety of questions," he explained. "There are two families of questions: about the user, their preference, type of smells, age, etc. All questions are closed ones, in order to make it easier for the user to reply (choose from multiple choices). The other family refers to the characteristics of the fragrances."
Gal adds, "After receiving the objective answers, the AI Wizard 'mixes' the answers, and based on these data chooses the most suitable items from the shop’s arsenal. Due to the subjective nature of fragrances, the system gives also three alternatives, which are similar to the most suitable one."
Finally, he says, "The key factor is to choose the right questions, so in one hand there won’t be too many of them, which will cause a boring process, and on the other hand to gain enough data to make the selection. This hurdle (finding the right questions and the right number of them) was resolved by hours of observations, study the nature of fragrances, and understanding the decision-making process."
As for the underlying data, Gal explains, "The relevant data is available either from the manufacturer, or from open literature. The data base grows according to more users, who come with new fragrances."
Yet a key question remains: How can personalization stay relevant without limiting users’ ability to discover unexpected fragrances?
"The whole idea is that the user doesn’t need to know the fragrance in advance," says Gal. "They just answer the questions, and the system introduces them with the correct match (which is available in the shop, or for online order)."
He concludes, "In reality, after 'smoothing' the questions, based on the observations and trials, the accuracy we’ve achieved is remarkable."
As retailers increasingly look to combine personalization with measurable commercial performance, the Sephora Marbella activation serves as a real-world demonstration of how AI-powered scent matching can reshape the fragrance shopping experience while introducing consumers to new launches and best-selling scents.
