Sephora Creates an Online “In-store” Experience through Generative AI

There’s no denying how much AI has developed and risen to fame. A global trend ignited by ChatGPT, the fame-rising chatbot system in November 2022, by OpenAI. Its adeptness in translating vast and heavily technical data into natural, human-tailored-like language has been raved worldwide, especially in the retail industry.

From content production, copywriting, and product renderings to personalized recommendations, platforms like ChatGPT does it all, sparking the potential for generative AI. Brands across various industries have started to notice and explore the possibilities of integrating its use into running the business.

Some of the many giant brands are leveraging this opportunity to streamline mundane tasks for optimized customer engagement, including:

• Carrefour’s experimenting with using ChatGPT and generative videos for FAQs.
• Shopify’s newest AI-based feature, Shopify Magic, automates the generation of product descriptions.

In particular, Sephora (Singapore) is using conversational AI to enhance its digital retail stores by blending in-store and online teams into an omni-retail department. Here’s how they’re doing just that.

Sephora: Evolving the digital retail experience with AI

Photo by Deva Darshan on Unsplash

The past three years created a shift in the “what” and “why” of retail stores as an influx of consumers shifted to digital stores, raising the question: “How do we incorporate the brick-and-mortar experience into our online stores?” Sephora took the leap by blending the physical and digital experience through conversational AI to offer par-level support and assistance, regardless of where their consumers shop.

Sephora’s analysis of the series of interactions its consumers took before purchasing, including product details, best prices, reviews, and recommendations, led to the introduction of its virtual assistant. Its role consists of the following:

Personalized consultations

Through conversational AI, Sephora created the bridge between customers and in-store Beauty Advisors on demand to create a personalized purchase roadmap by providing customers with product knowledge, making recommendations, and reserving and picking up their orders at their convenience. Hence, reducing the friction of purchase resistance.

Bringing the “in-store” experience online

Sephora’s virtual assistant lets customers reserve products, pick them up in-store, check product availability, and ask questions about store timings and return policies. The beauty advisors in-store answer any questions the customer may have about their purchase decisions as they would when they're in-store.

This led to over 9,000 users on the bot (6,000 in Singapore and 3,000 in Malaysia) within a year, over 332,000 conversation sessions (250,000+ in Singapore and 82,000 in Malaysia), and a $30,000 average incremental monthly revenue.

“Chatbot has been looked at primarily for resolving customer complaints, but right now, we are not using it to upsell and cross-sell, and we could actually use this to figure out their purchase. Few companies do attempt it, so this remains a hot topic for discussion or look at on these platforms. The success of commerce in chatbot has not been discussed much.” - Gaurav Khanna, Senior Lead, Data Science - Recommendations, Personalization, and Chatbot, Tokopedia. 

Key usage considerations

One of the biggest challenges with generative AI is bias, as it is based on human-generated data by using an algorithm that prioritizes certain types of content over others. For instance, if one asks a virtual assistant for a lipstick recommendation, its answer may be influenced by whatever is inputted into its database, which may not necessarily be the best option.

Another limitation is environmental impact; the tool requires an internet connection and generates hefty content for each user request, consuming large amounts of power and bandwidth when used by many users. Given the gradual pressure over corporate social responsibility, companies must consider its environmental implications or, better yet, find ways to mitigate the negative impacts.

Moving forward with generative AI 

The potential and possibilities of leveraging the power of generative AI only gets more exciting here as the world seeks to learn more about how the technology works and how to optimize it to its full potential. Sephora’s utilization of generative AI as a virtual assistant proves that it can be implemented successfully.

However, generative models often require large amounts of data to work well; creating new content from scratch will take some time to tweak it to provide the functionality needed. There is still a significant knowledge gap that needs further exploration and learning, given the maturity of the technology.

With the dynamism of the consumer market, there is a need to keep up with that pace. Generative AI can be a critical tool for companies to create opportunities and establish a first-mover advantage to meet the consumer's needs. Though Sephora’s approach toward blending the online and in-store experience is a great start, AI can never replace human-to-human interaction.

Companies must discern where machines and humans would create the most value in the purchasing journey. Chinese e-commerce giant, JD.com, is spearheading this approach as it plans to launch ChatJD, their “industrial version” of ChatGPT, primarily focusing on retail and finance by generating content from large datasets and having a human-to-computer dialogue.

An approach that many others still have to trail behind to keep pace with technology and its utility.