Our Methodology
How we evaluate and recommend fragrances at Lavender Thorne.
Data Sources
Our fragrance profiles are built from multiple sources:
- Community Reviews: We analyse thousands of user reviews to understand how fragrances perform in real-world conditions—on different skin types, in various climates, and across seasons.
- Official Fragrance Data: We include manufacturer-provided notes, accords, and composition details as the foundation of each profile.
- Performance Metrics: Longevity, sillage, and projection data comes from aggregated user feedback, giving you realistic expectations.
How We Evaluate Fragrances
Each fragrance is assessed across multiple dimensions:
Scent Profile
We categorise fragrances by their dominant accords (woody, floral, fresh, etc.) and detail their note progression from opening through dry-down. This helps you understand not just what a fragrance smells like initially, but how it evolves over hours of wear.
Performance
- Longevity: How long the fragrance remains detectable on skin
- Sillage: How far the fragrance projects from your body
- Projection: The "bubble" of scent around you in the first few hours
Context
- Seasonality: When the fragrance works best (spring, summer, autumn, winter)
- Occasion: Where it fits (office, evening, casual, special events)
- Gender: Community perception of masculine/feminine/unisex appeal
Value
We consider price relative to quality, longevity, and alternatives. A £20 fragrance that performs well can be a better recommendation than a £200 one that underdelivers.
AI-Generated Descriptions
Many of our fragrance descriptions are generated using AI trained on community reviews. This approach allows us to synthesise insights from hundreds or thousands of user experiences into cohesive, readable profiles.
Our AI is instructed to write in an honest, conversational tone—highlighting both strengths and weaknesses. We don't use marketing language or manufacturer talking points. If a fragrance is polarising, overhyped, or has significant drawbacks, the description will say so.
Editorial Content
Our blog posts, celebrity fragrance guides, and curated collections are written by humans with genuine passion for perfumery. These represent our editorial perspective and are clearly distinguished from AI-generated content.
What We Don't Do
- We don't accept payment from brands for featured placements
- We don't copy manufacturer marketing descriptions
- We don't inflate ratings to drive affiliate sales
- We don't recommend fragrances we wouldn't genuinely suggest to a friend
Continuous Improvement
Our methodology evolves as we gather more data and feedback. If you notice inaccuracies or have suggestions for improvement, please let us know.