Two customers can love the same perfume for different reasons. One loves the citrus opening, the other the creamy sandalwood base. A personal scent profile captures those preferences so that recommendations get better over time.
What a scent profile contains
- Liked and disliked accords: for example, loves amber and rose, dislikes leather and heavy sweetness.
- Intensity preference: skin scent or statement projection.
- Occasions and seasons: office, evening, weddings, summer.
- Reference perfumes the person owns or wants.
How AI builds it
Every search, click, sample request and “yes, that's right” confirmation is a signal. AI weighs these signals to estimate which accords matter most to the person, then ranks new oils by how well they fit. Crucially, explicit feedback (“this is correct”, “not for me”) is far more valuable than guesses from browsing alone.
Why this helps businesses
For brands and retailers, anonymous, aggregated profiles show which accords are rising in demand and which combinations customers ask for but cannot find, a direct input into new product development.
Privacy first
Personal profiles should be transparent and controllable: customers should know what is stored and be able to clear it. At Haveli AI, the Perfume Decoder stores its memory in the customer's own browser, and we do not store photos.