blog
AI provide faster responses to Matas skin test
Matas offers all customers an online skin test to help them provide the best advice. Behind the scenes, Matas' experts are at work, and now they are getting extra help from AI, which provides faster responses.
Offering the best and most relevant customer experience is crucial for Matas, whether it's in-store or online. One of their most popular initiatives is the opportunity to take a skin test online. Here, customers can answer a series of questions, which are saved on their profile. Behind the screen, the answers are analyzed by Matas’ beauty specialists. This analysis then forms the basis for personal product recommendations both online and when the customer receives advice in-store.
The beauty experts manage the analysis of the skin test and the selection of personal product suggestions, but they are now being successfully assisted by a new AI model, so that each expert can respond to more customers faster.
"We see great opportunities in the use of AI. Especially with processes where we do the same thing over and over again, but we will not compromise on the quality of our advice, so we closely monitor artificial intelligence. Our beauty experts have continuously evaluated the quality of the AI model's suggestions, and it now receives an average rating of 4.6 out of 5. With such high quality, we see significant time savings. This is just one of a number of ongoing AI projects, and several other initiatives are on the way with equally promising results, so it's an exciting time."
Brian Andersen, Group Digital Director, Matas
Strong AI model built into the e-commerce solution
The AI model's knowledge about skin tests is built by our solution partner Hesehus based on 33,000 historical responses written by Matas' experts. The suggestions from artificial intelligence are built directly into the e-commerce solution. This means they are automatically suggested in the sales assistants' interface, making their daily life easier.
The first version went live in February, but since September it has really made a difference, where retraining was incorporated. It means the model includes the new skin test responses every week, and at the same time, it also considers the sales assistants' feedback for each recommendation. It continuously learns from both the good and not-so-good recommendations.
Questions?
Our team is ready to help
