The image states “60% of customer support cases deflected.” To the right is a chat interaction on an iPhone of Pandora’s customer service agent Clara assisting a customer with her order status.
Pandora logo

Pandora makes every online shopper feel like a gem with Agentforce.

AI agents help customers find the perfect jewelry and get instant answers to their questions.

Summary

Pandora wants to replicate the magic of in-store shopping online. With Agentforce, they’re focusing on building personalized recommendations and on-demand guidance to boost customer engagement and loyalty.

About

Pandora employs 37,000 people worldwide to design, manufacture, and sell hand-finished jewelry made from high-quality, 100% recycled silver and gold at accessible price points.

The Results

60 %
autonomous case deflection
10 %
boost in net promoter score with agent-first service
22 %
of total sales handled through Commerce Cloud

Pandora’s bold growth ambitions required a reimagined digital shopping and customer support experience.

Pandora began in 1982 in a small Copenhagen shop, where a jeweler and the brand’s first in-house designer set out to create unique, hand-finished pieces. Today, Pandora is the world’s largest jewelry brand, and their charms, bracelets, rings, and necklaces are sold in more than 100 countries through 6,800 points of sale, including 2,700 concept stores.

“What sets us apart is that our jewelry is something you can associate with specific memories or passions,” said Baltazar Hasselsteen Ozonek, Vice President of AI and Innovation for Pandora. “Our store associates are specially trained to help customers connect their stories and moments to an item that will hold meaning forever.”

For decades, physical stores where associates built personal connections with customers and shared the stories behind each piece have driven Pandora's growth. This storytelling is instrumental in helping customers choose meaningful gifts for loved ones, and Pandora saw an opportunity to extend their in-store magic further, into the digital world.

“You could spend an hour in-store with one of our wonderful associates putting together a bracelet for your loved one and talking about your family,” said David Walmsley, Chief Digital and Technology Officer. “That richness of conversation is the heart of the Pandora experience and always will be. We want to figure out how to bring that online.”

Part of creating that seamless digital journey also meant reimagining customer service. In 2024, the company’s team of 500–600 specialists managed a considerable volume of inquiries via email, web, and live chat. These specialists are dedicated to providing personalized service, but manually responding to each inquiry could slow response times, leading to longer service queues. Even though Pandora brought in seasonal support staff to support holiday surges, peak demand still overwhelmed capacity. A traditional chatbot was able to assist with “Where is my order?” requests, which account for a significant amount of their total service inquiries, but the chatbot’s rigid, transactional conversation style didn’t match Pandora’s trademark warm, personalized service.

To reach their goals around providing even better online experiences, Pandora knew they needed a way to automate routine requests and create a digital storefront that captures the same magic customers feel in person.

That richness of conversation is the heart of the Pandora experience and always will be. We want to figure out how to bring that online.

David Walmsley
Chief Digital & Technology Officer, Pandora

Agentforce is expected to increase online conversion rates with personalized recommendations. 

Pandora is using Agentforce, the agentic layer of the Salesforce Platform, to make online shopping feel as personalized and thoughtful as stepping into a store. With Agentforce, they built a personal shopper agent named Gemma that’s designed to help customers choose the perfect piece of jewelry. Gemma gets to know what each shopper is looking for and then makes recommendations based on the occasion, who the gift is for, and the shopper’s budget.

For example, imagine a customer visits Pandora’s website looking for a gift for their mother. They start a chat with Gemma, which greets them and asks a few simple questions to understand who they’re shopping for and what their budget is. When they share that their mother loves ballet, Gemma surfaces pieces inspired by dance, movement, and elegance, weaving in anecdotes and product stories much like an associate would in-store. The customer feels guided, understood, and supported, while discovering items they may not have found on their own.

“For the first time ever, we’re going to be able to have a dialogue with shoppers online,” said Ozonek. “We’re creating something that will set us apart by taking the best of that in-store experience and bringing it online.”

To deliver tailored recommendations, Agentforce draws on data across Pandora’s ecosystem. It reasons across customer order history and product information from Commerce Cloud, as well as recommendation data from Bloomreach and unstructured insights like mystery shopping and UX research insights stored in Databricks via Data Cloud. Together, this data helps Agentforce pinpoint key moments in customer interactions, understand context, recognize when someone is ready to purchase, spot cross-selling opportunities, and deliver more thoughtful, timely recommendations. And while the qualitative research guides Gemma’s sales approach, granular tone controls ensure it sounds like a Pandora sales associate by matching the warm, distinctive tone used by in-store associates.

Pandora is testing Gemma in select markets before rolling out globally, but they’re excited about the expected impact. With Data Cloud, they can analyze conversations that once disappeared in brick-and-mortar stores — like which questions customers ask most often, which stories and recommendations resonate, and which tactics drive conversions. These insights will fuel future marketing and sales campaigns.

“I want to create an overview that will literally become the world’s biggest library of people’s loves,” said Ozonek. “This will inform the next decision, the next experience, and our next value proposition as a brand.”

We’re creating something that will set us apart by taking the best of that in-store experience and bringing it online.

Baltazar Hasselsteen Ozonek
VP, AI & Innovation, Pandora

Agent-first service answers FAQs instantly and provides on-demand order status updates.

To ensure that thoughtful, personalized experiences continue after each sale, Pandora built a customer service agent named Clara to automate routine customer service requests. Built in collaboration with Salesforce Professional Services and Publicis Sapient in just weeks, Clara specializes in answering “Where is my order?” requests and FAQs like how to clean and store jewelry.

Instead of rigid, transactional chatbot conversations, Clara uses friendly, conversational language to help customers quickly. For example, when a customer wants to track their order, they can start a chat with Clara and type, “I ordered stacking rings last week. When will they arrive?”

Clara requests an order number or email address, taps into order information from Service Cloud and Commerce Cloud, and responds in seconds.Commerce Cloud brings in fulfillment details from Pandora’s IBM Sterling order management system via direct connectors, giving Clara real-time visibility into order status.

If a request goes beyond what Clara is designed to handle, such as changing an order after it’s been placed, Clara seamlessly hands the case to a human representative. Clara always includes a summary of each conversation so representatives get up to speed quickly and customers don’t have to repeat themselves.

“We were blown away by Agentforce right away,” said Walmsley. “You get the cellophane off the software, get it installed, and it’s like, okay, that actually works.”

Clara is already deflecting 60% of cases, and a net promoter score increase of 10% indicates that customers are more satisfied with the service they’re receiving. Pandora started by routing one-tenth of their FAQ traffic in North America and the United Kingdom to Clara and are scaling with more traffic and more markets as their confidence in Clara grows.

For now, Clara is separate from Gemma, but Pandora sees Agentforce as a foundational element for their future digital shopping experience. As they transform into an agentic enterprise, they envision a seamless digital experience where customers get assistance from AI agents before, during, and after each sale — wherever and whenever they need it.

We were blown away by Agentforce right away. You get the cellophane off the software, get it installed, and it’s like, okay, that actually works.

David Walmsley
Chief Digital & Technology Officer, Pandora

The Salesforce Platform powers ecommerce and loyalty across global markets for Pandora.

Pandora is driving growth through digital sales with a sophisticated and reliable ecommerce experience built on the deeply unified Salesforce Platform. Their Salesforce apps work hand in hand with third-party solutions to deliver the personalized shopping and service experiences customers expect.

“Salesforce could have gone down the route of saying, ‘No, we’re a walled garden. Take our products as they are.’ Instead, they pivoted toward composability,” said Walmsley. “All the work Salesforce has done to open up the platform has made it much easier to work with in a modern engineering sense.”

Commerce Cloud powers Pandora’s ecommerce business across the U.S., Canada, Europe, and Asia-Pacific region — with Latin America coming next.Commerce Cloud handles 22% of total sales and provides a scalable platform for Pandora’s global ecommerce operations. They began their ecommerce journey with Salesforce’s digital storefront reference architecture and are now moving toward a fully composable setup with the PWA Kit for faster, app-like shopping across devices. For search and personalization, they take a best-of-breed approach.

Behind the scenes, MuleSoft connects Commerce Cloud with IBM Sterling for order management, while Kong manages corporate APIs. A digital orchestration and service layer ties it all together, helping teams across ecommerce, stores, and customer service create connected shopper experiences. Built on a microservice, API-led architecture, the system is flexible and easy to adapt — simplifying integrations, speeding up customizations, and making it easier to keep pace with changing customer needs.

This approach shines during peak shopping moments. In the 2024 holiday season, Pandora saw an increase in orders, more site traffic, and a lift in average order value compared to the previous year. AI-powered recommendations from Einstein also boosted products added to carts, which drove a jump in attributed revenue.

On the service side, Pandora replaced Zendesk with Service Cloud to support B2B and B2C customers. Now, reps have a complete view of each shopper’s history, making support faster and more personal.

Looking ahead, Pandora is extending that same personalization to loyalty. Working with Salesforce Professional Services and Publicis Sapient, the company recently launched Loyalty Management in Austria, Germany, France, the U.K., the U.S., and Canada. Pandora’s loyalty program has millions of dedicated members worldwide and Loyalty Management helps Pandora reward them with personalized promotions that inspire long-term brand advocacy.

All the work Salesforce has done to open up the platform has made it much easier to work with in a modern engineering sense.

David Walmsley
Chief Digital & Technology Officer, Pandora

With Salesforce, Pandora brings its ecommerce, customer service, and loyalty data together on a deeply unified platform. “I love that I don’t have to focus so much on the infrastructure and the hurdles of managing the infrastructure,” said Catarina Runa Miranda, Global AI Director and London Digital Hub Lead. “Without that pain, I can focus way more on coming up with innovative new ways to capture the in-store experience online.”

But innovation only works if customers can trust it. The Salesforce Trust Layer, along with built-in governance controls, were key to Pandora’s decision to partner with Salesforce for their agentic AI. Agentforce’s zero-retention capabilities that automatically delete prompt outputs after every interaction keep Pandora compliant with European data retention rules. With these safeguards in place, Pandora felt confident rolling out AI agents broadly, balancing innovation with privacy and compliance.

Ease of use has been another driver of adoption. Low-code and no-code Agent Builder tools make it simple to design and launch agents quickly. “Agentforce doesn't require a highly technical person to operate,” said Ozonek. “It’s configurable by anyone, which means our nontechnical staff who understand each aspect of a business challenge can build agents to solve those challenges.”

With their customer-obsessed commerce, unified platform, and strong privacy safeguards, Pandora is setting a bold new standard for the future of retail.

I love that I don’t have to focus so much on the infrastructure and the hurdles of managing the infrastructure.

Catarina Runa Miranda
Global AI Director & London Digital Hub Lead, Pandora