When the Austrian crystal company Swarovski embarked on its implementation of generative AI (GenAI) at scale, it focused on building a robust data governance framework, training more than 10,000 of its staff members, and balancing speed with responsible implementation. Speaking to the Luxury 2050 Forum, IMD’s community of leaders focused on innovation in the luxury sector, Swarovski’s Chief Digital and Information Officer, Lea Sonderegger, outlined how the company prioritized a focus on the human side of the implementation, placing a strong focus on the impact it would have on employees. She also discussed the big-picture challenges of scaling from pilots to production.
When Swarovski piloted one of its first AI use cases, which focused on Customer Relationship Management (CRM), it avoided treating the project simply as a productivity shortcut. Instead, the team used GenAI to shift from customer communications with broad messaging aimed at targeted segments to hyper-personalization. Each customer received communications tailored to their individual style, preferences, and purchase history.
In practice, this meant that individuals received styling suggestions or recommendations to enhance their shopping experience. The result is that personalization has become a central part of customer experience. The initiative delivered immediate revenue uplift through higher conversion rates, while creating strategic momentum by demonstrating return on investment.
Central to the success of the initiative was the CEO’s support for the project, Sonderegger said. “If you don’t have the backing of your CEO, it’s mission impossible,” she noted, explaining that the Swarovski leadership’s behavior enabled cross-functional collaboration and dialogue across different parts of the organization, which helped to address risks and identify opportunities.
By systematically emphasizing a value-first approach rather than technology-driven implementation, Swarovski achieved more than 5% contribution to digital commerce with less than 1% of digital budget. The company takes a disciplined approach to tracking the value of its AI projects. Before starting, the company assesses the estimated impact – such as increased revenue, lower costs, or improved productivity. After launching, it tracks the actual results using a live dashboard. This continuous monitoring gives Swarovski real-time visibility on the value that is being created. It keeps things transparent across all projects and allows the company to shift investments towards more successful initiatives, while quickly identifying and addressing those that aren’t delivering as expected.
The Swarovski team benefited from a long-term focus on data governance even before AI entered the conversation.
The Swarovski team benefited from a long-term focus on data governance even before AI entered the conversation. The team has been building a data lake to store all of the company’s proprietary information in one place. As a result of that investment, the company can provide a seamless omnichannel personalized customer journey. Sonderegger noted that in any GenAI project or initiative, governance will slow down transformation, but that it is mission-critical to balance governance, working with compliance and legal teams, along with innovation.
She explained: “If the data is not correct, you will not get intelligent insights or intelligent recommendations. And if you make that tangible to the people that work with it, suddenly they see value in putting effort into this.”
One way to build a strong data policy is to think of data as one of the company’s products. Traditionally, when we consider products, we consider the capital and labor required to produce the item. This framing encourages leaders to accept responsibility for the role they play in data production and maintenance. They will think, “I need to manage this data, because otherwise I am leaving money on the table.”
Transform your career and business with next-generation AI and digital skills
Rather than simply directing employees to use GenAI to complete their work more quickly, the sell to Swarovski’s more than 18,000 employees, with operations in 140 countries, was that the GenAI adoption would allow them to do work that previously had been out of reach. The technology amplifies and scales professional judgment rather than replacing it, and the leadership team’s communication and actions were consistent with this idea. As Sonderegger put it: “AI will never replace human creativity. It can just support human creativity, even supercharge human creativity.”
Sonderegger noted that one use case had failed initially: an attempt to create 3D designs from 2D sketches. By working closely with the design team, they were able to determine that the technology was not yet up to the task. However, she noted that as technology continues to advance, they may revisit that use case. Sonderegger said: “It’s a journey, but it’s clear for us that we are a brand where human design and creativity must stay at the center. That’s our essence.”
Working with the HR department was critical to widespread adoption. Employee trust grew incrementally, earned through visible improvements in day-to-day work. Employees could make decisions guided by evidence rather than hierarchy, reinforcing their sense of ownership and professional identity. Scaling this transformation required investment not only in platforms, but in people.
Swarovski embedded 50 AI “champions” or “ambassadors” across business functions to support colleagues, acting as local enablers rather than technical experts. These self-nominated champions leveraged training from large tech companies including Microsoft, Google, or AWS, and then the team used these champions as force multipliers across their various functions. The role of the champions was to support colleagues in their work, experiment with tools such as Microsoft 365 Copilot, identify problems the AI could help solve, and act as a bridge between their function and the central data and AI team.
Training was designed in tiers. All employees completed role‑appropriate AI literacy training, while AI champions received additional instruction through onboarding materials, early access to AI tools, and peer exchanges. Peer dialogue and experimentation were core to the model, which the leadership team had heard from other companies was one of the most effective ways to drive adoption. AI champions shared examples and lessons learned, while live demonstrations in community channels and quarterly exchanges ensured AI capability spread organically.
Another common trap for GenAI pilots is the maintenance trap when it comes time to scale.
Another common trap for GenAI pilots is the maintenance trap when it comes time to scale. Many organizations start building something that is relatively narrow in scope, addresses a few use cases, and has few interdependencies with other systems. However, when the team sees success with the pilot and wants to scale, the product (often internal agents) does not suit broader applications because of the maintenance required. Companies can avoid this trap by keeping the focus of the particular product relatively narrow, without many interdependencies.
In this context of the scope of a particular agent, Sonderegger addressed the buy vs build decision that teams must make. She noted that there are an increasing number of tools with built-in AI solutions, allowing companies to quickly generate insights. But she said that, for Swarovski, one of their most pressing challenges was demand planning, for which the team needed a bespoke solution. She noted that AI “only knows what it knows,” but “It’s not good at predicting the future if there is no pattern.” Swarovski’s next steps, she said, are to focus on agents across three areas: first, an enterprise knowledge concierge; second, an analyst that retrieves data from the CRM database about customer behavior; third, an image-based agent that works from photographs of a store shelf to report on productivity and propose merchandising guidelines.
1. Normalize the use of GenAI by highlighting the process. If you’ve generated a report or analysis, don’t just present the output to your team; take them through how you got it, highlighting how AI is a “co-thinker” but not the ultimate decision-maker.
2. Start with the value, not the tech hype. Focus on why you are embarking on the project, not how.
3. Get your data house in order before you scale. This requires a strong data governance framework.
4. Ensure your human and tech investments are equal. You won’t succeed without your people.
AI can optimize processes, but true leadership requires human judgment, says Shelley Zalis. Success depends on integrating machine intelligence with ‘soft skills’ like empathy and intention....
Audio available
Facing a fundamental redefinition of their purpose in organizations, today’s CIOs need to pivot quickly. ...
Hugo Travers built a 20M-strong news audience by prioritizing accessibility, trust, and transparency, offering leadership lessons for the digital age....
To unlock true competitive advantage, leaders must learn to integrate the refined intelligence of American AI models with the coordinating power of China. Howard Yu explains why most Western companies are on...
Audio availableExplore first person business intelligence from top minds curated for a global executive audience