A luxury beauty brand wanted us to come up with a recommendation engine for fragrances to target customers with perfumes with personalised recommendations based upon behaviour and personality, to drive them to buy a new perfumes that they were more likely to buy.
Our data science team used machine learning to formulate 18 simple, effective questions for our client’s customers to build their individual preferences. The data collected gave us the raw material to: Cluster the population into 12 personas, defined rating rules to quantify how much a customer likes a product, we designed five recommendation engines to work in symbiosis and make recommendations
Testing is critical to validate your recommendation engine, and we did this with the public at Luton airport and on Oxford Street. 82.3% of respondents indicated that their recommendations were relevant, giving high scores of 4/5. Taking this user insight, we further refined the engine and put it into production.