A new retail battleground is emerging and it’s the AI assistant deciding what to recommend before a shopper ever visits either one. That makes loyalty more valuable, not less.
Emmy Brown Berlind, SVP and general manager of loyalty at Sephora, joins Matt Britton to explore how a nearly 20-year-old loyalty program is adapting to that shift.
Their conversation covers why the Beauty Insider program is designed to feel personal instead of transactional, and how Sephora uses AI without crossing into “creepy.”
She also discusses how its combination of first-party data, beauty advisors, stores, and emerging agentic shopping partnerships could help the brand stay relevant wherever beauty discovery moves next.
About the guest:
Emmy Brown Berlind is SVP and general manager of loyalty at Sephora, where she leads the Beauty Insider program serving more than 46 million members. She has spent more than a decade at Sephora across innovation, content, commerce, and loyalty, giving her a broad view of how the brand connects stores, digital experiences, rewards, creators, personalization, and emerging AI-powered tools.
Before Sephora, Emmy worked at Clorox and Yes To, experiences she credits with helping her understand brands at different stages and scales.
Key Takeaways:
[02:47] When Experience is at Stake, Stores Still Win — Sephora’s physical stores remain central because the experience extends beyond access to products. Emmy points to highly trained beauty advisors, discovery-oriented store layouts, and sample rewards that let loyalty members try something new. The lesson for retailers and marketers alike is that digital convenience does not automatically replace physical retail when the store creates expertise, exploration, and value that online channels struggle to reproduce.
[06:07] Loyalty Has to Feel Personal, Not Transactional — As Beauty Insider approaches its 20th anniversary, Emmy says Sephora has kept evolving the program around choice. The birthday gift has expanded from a simple reward into multiple options, including points, because clients value choosing for themselves. That philosophy reflects the broader goal: loyalty should make customers feel recognized rather than simply rewarded for spending again.
[07:36] Personalization Can Turn Rewards Into Discovery — Beauty Insider Challenges use limited-time tasks and point rewards to encourage members to try something new. The tasks can be personalized based on purchase history. For example, prompting someone who has never bought a hair product at Sephora to explore that category. More than 30% of clients have participated in a challenge since the format launched in 2023.
[09:09] AI Shouldn’t Feel Creepy — More than 90% of Sephora’s sales come from Beauty Insider members, giving the brand a rich first-party data set. Emmy’s standard for using that data is straightforward: apply AI where it creates practical value for the client. Tools such as Beauty Skin Scan can improve shade and skincare recommendations, while the AI Beauty Chat remains optional. The technology earns its place by making the experience easier, not by showing off what it knows.
[14:31] Follow the Shopper Into the Agentic Era — Sephora does not expect every beauty journey to begin on Sephora.com. Emmy says the brand wants to be wherever people are shopping for beauty, which now includes AI-powered environments. Sephora has an app inside ChatGPT and a Google Agentic Checkout partnership that lets consumers discover and purchase without leaving those platforms. The strategy is less about protecting a destination and more about protecting relevance.
[18:12] Client-Friendly Decisions Are Not Always the Easy Ones — Emmy’s leadership philosophy comes back to a question she encountered early at Sephora: “Is this client friendly?” The answer can create additional operational complexity. Birthday gifts, rewards, and loyalty experiences require coordination across multiple departments, but Emmy argues that advocating for what clients genuinely want is part of the job, even when the simpler internal answer would be easier to execute.
