L'Oréal handles consumer support cases 64% faster with Agentforce

Meeting rising expectations

L'Oréal is the world's leading beauty company, with dozens of global brands selling directly to consumers through a network of brand websites. When consumers reach out, they expect natural, conversational chat support on demand at any time of day. 

For L'Oréal, simply adding a chatbot wasn't an option. Every interaction had to reflect each brand's distinct voice, and when a conversation required an advisor, the handoff needed to be seamless.

Behind the scenes, the technical landscape was equally complex. Separate order management systems supported the US and Canada, consumer records lived in third-party systems, and each brand had unique requirements. L'Oréal needed a flexible solution that could support those complexities while delivering a consistent consumer experience.

More than a chatbot

To deliver the seamless, brand-aligned experience they envisioned, L'Oréal launched an order status agent powered by Agentforce on YSLbeauty.com and Kiehl’s.com. Available 24/7 through live chat, it provides instant order status updates and brings an advisor into the conversation when needed.

Imagine a shopper checking on a recent order late at night. They open the chat and ask, "Where's my order?" — and after verifying their identity, Agentforce retrieves the latest order and tracking information in seconds. Because one agent framework serves both sites with configurable subflows adapting tone, wording, and approved responses to each brand and market, that same capability sounds like YSL on YSL Beauty and like Kiehl's on Kiehl's.

If a shopper needs additional help or a question falls outside the agent's scope, Agentforce transfers the conversation to a consumer care advisor with the entire chat history attached. During business hours, it routes to the appropriate care team, while after hours the shopper is told an advisor will follow up within one business day.

More than a chatbot

Imagine a shopper checking on a recent order late at night. They open the chat and ask, "Where's my order?" — and after verifying their identity, Agentforce retrieves the latest order and tracking information in seconds. Because one agent framework serves both sites with configurable subflows adapting tone, wording, and approved responses to each brand and market, that same capability sounds like YSL on YSL Beauty and like Kiehl's on Kiehl's.

If a shopper needs additional help or a question falls outside the agent's scope, Agentforce transfers the conversation to a consumer care advisor with the entire chat history attached. During business hours, it routes to the appropriate care team, while after hours the shopper is told an advisor will follow up within one business day.

Grounded in real-time order data

When a shopper requests an order update, Agentforce verifies their identity and retrieves order data through custom APIs built on L'Oréal's existing integration services. The US and Canada run on different order management systems, so every transactional action branches on country and brand to call the right one, then returns tracking details in the same conversation. Consumer records come through external service callouts that confirm the purchase before any order detail is shared.

The Atlas Reasoning Engine determines the next best action for each conversation. It can answer order status questions, respond to product or promotion inquiries, or escalate the conversation when human support is needed. Escalated chats route through Service Cloud on L'Oréal's existing skills and queues, which resolve by brand and country. When no advisor is available, Agentforce creates a case, gives the shopper a case number, and confirms by email.

These are products people put on their skin, so L'Oréal constrained what the agent is able to say. Responses come from a library of pre-approved, translatable language rather than sentences the model composes on its own, and anything edging toward a medical or otherwise sensitive topic routes straight to an advisor. The reasoning engine chooses which approved response applies; it never writes a new one. For a house of luxury and premium brands, that limit is the point. The constraint is what made a consumer-facing agent approvable in the first place.

A shared global layer holds the conversation structure, response logic, and guardrails, while subflows specific to brand, country, and order management let each market tailor tone, wording, and approved responses. Brand-level details live in a Salesforce custom object: name, code, welcome message, and the resources each brand points consumers to. Bringing on a brand becomes a configuration change rather than a new build, which is how one agent framework covers a multi-brand, multi-market rollout without a separate deployment for each.

Built for expansion

L'Oréal intentionally started with a single consumer use case before expanding more broadly. The initial goal is to autonomously resolve 60% of order inquiries, which will reduce the need for consumers to contact an advisor while providing immediate answers around the clock.

The first deployments on YSL Beauty and Kiehl's are allowing L'Oréal to validate performance, refine conversations, and measure business outcomes before introducing the experience to additional brands.

With the core architecture now in production, L'Oréal plans to expand the experience to additional brands, with Lancôme web chat and voice experiences next on the roadmap. Future releases will also extend the agent's capabilities to answer FAQ and knowledge-based questions, building on the same architecture already supporting live consumer conversations.