Ometria Labs: Feelunique
Europe’s largest online beauty brand uses machine learning to personalise based on brand affinity.
At Ometria, we know our most significant product milestones have been achieved by closely collaborating with our amazing clients. Ometria Labs is our innovation stream, where we test out new and innovative ideas for potential features that solve retail marketers’ biggest day-to-day challenges, and share the results with you.
Feelunique is Europe’s largest online beauty retailer, and is passionate about offering customers innovative, digital-first experiences.
With a range spanning 30,000 products from over 500 established and emerging brands, the Feelunique marketing team wanted to make sure that customers were only targeted with products and brands that were most relevant to their interests and tastes.
To achieve this, the team used Ometria to segment its newsletter audiences, based on the brands that customer had previously purchased from.
While this produced greater engagement and higher revenue per email compared to non-segmented, unpersonalised sends, Feelunique wanted to use a more sophisticated approach to identifying shopper tastes and preferences that took into account factors beyond simply what a customer had previously purchased.
So the marketing team used Ometria’s AI-based Predictive Segmentation functionality to build its newsletter audiences.
Using Ometria’s predictive segmentation engine, the Feelunique marketing team sent subscribers a personalised newsletter campaign that targeted shoppers based on their affinity towards certain brands.
Each of Feelunique’s newsletter subscribers received a version of the campaign that was themed around a cosmetics brand that was most likely to interest them.
Audiences were automatically created using machine learning, rather than relying on simple rule-based segmentation.
The taste profiling algorithm takes into account factors such as:
- The individual’s purchase and browsing behaviour.
- Relationships between similar people and products.
- Temporal correlations, based on the buying cycle of each product.
- Reinforcement learning, based on whether a shopper responds to machine-generated predictions in the expected way.
So who won this test of human- versus machine-based segmentation?
Compared to the version that relied on simple rule-based segmentation, the AI-segmented email achieved a:
- 95% uplift in revenue per email
- 33% increase in click to open rate
- 12% increase in average order value