SEPHORA BEAUTY BAE - GEN AI CHATBOT
INTRO
Every day, thousands of customers visit our brick-and-mortar stores, seeking personalised beauty services from our expert Beauty Advisors. With Beauty Bae, we're bringing that same level of tailored, expert beauty guidance right to our users' fingertips through our app.
​
Now, customers can effortlessly ask beauty questions in a natural, conversational way and receive personalised advice, including product recommendations, in an engaging and secure environment.
PROOF OF CONCEPT PHASE
A quick POC was implemented to test the viability of Beauty Bae. We decided to introduce the chatbot during product discovery of the purchase journey funnel. The entry point consist of the Beauty Bae icon stickied to the bottom right corner of the listing page. GIF was used to draw the attention of the user to the icon.
INITIAL RESULTS AND FINDINGS
The numbers were promising in the initial pilot, with a conversion rate of 12.59% (the app has a typical CR of 2-3%). Of users that engaged with the chatbot, there's a 47% click through rate to a product recommended by Beauty Bae.
​
However, traffic and usage of Beauty Bae is low as users do not engage with the feature, with only 2.4% clicking on the icon and only 50% agreed to the Terms Of Use.
PHASE 2: CONTEXTUAL ENTRY POINT
To address low usage of a feature that otherwise has great potential, we pivoted by introducing Beauty Bae in a more contextual and seamless manner.
We target users that can do with the help of the chatbot by introducing it after a certain amount of scroll depth. Users that are browsing the listing page are generally exploring options and would welcome the aid of Beauty Bae.
​
The suggested prompts in the chat screen are also contextualised to the listing category that the user is browsing on.
PHASE 2: MORE USE CASES
We also introduced Beauty Bae in the Product Description Page, specifically on the description and reviews section.
​
With a single tap, users send a pre-generated prompt to the chatbot to ask about product suitability and reviews summary. This helps them make a better and informed purchase decision.
FINAL: USER JOURNEY BASED AI INTEGRATION
Based on the learnings so far, we find that the barrier to usage is the initiation of the feature itself. From a user journey perspective, which has already been UX optimised, the introduction of the chatbot is seen as a detour or friction.
Rethinking about the value of AI, which is to help humans automate repetitive tasks and improve decision making, we went back to the drawing board, this time applying a greater emphasis on the user experience of the AI.
​
​Being a LLM, Sephora Beauty Bae is best suited to perform natural language processing tasks, meaning information-rich sections such as the product description or product reviews are greats areas of opportunity for the AI to help users summarise and give recommendations.
​
Instead of having to manually trigger the prompt in order for Sephora Beauty Bae to work, integrating it directly into the product description page with the output already generated is much more ideal for the user.
PRODUCT DESCRIPTION
Upon landing on the page, the product description summary is instantly generated. The user’s unique beauty profile is analysed and matched with the product's key features, allowing Beauty Bae to dynamically generate a personalised summary that highlights whether the product is a good fit for the user based on their preferences, skin type and skin concerns, helping users make informed decisions more efficiently.
​
Under the Highlights section, a suitability rating is generated to quickly inform users about their compatibility with the product.
PRODUCT REVIEW
Under the reviews section, Beauty Bae digests and gives a summary of the reviews of the product. Key words about the product is picked out for the users to get a quick overview of what to expect if they were to buy it.