As one of the world’s oldest and most innovative materials companies, Saint-Gobain operates in over 70 countries and serves diverse sectors including construction, mobility, and industrial markets. With a legacy dating back over 350 years, the French multinational continues to evolve, placing data and AI at the heart of its digital transformation strategy.

In this third and final part of our interview series, Benoit Lepetit, Group Chief Data & Analytics Officer at Saint-Gobain, sits down with Julian Schirmer, Co-Founder of OAO, to discuss what the company has learned so far, what has worked, what hasn’t, and how they’re building organizational muscle for the future.

In part 1, Lepetit explored the why behind Saint-Gobain’s data and AI ambitions, emphasizing the “three Cs”: customers, collaboration, and competition—with sustainability as a fourth strategic pillar. Part 2 focused on the how—touching on federated governance, transformation enablers, and creating strong internal capabilities.

Now, in part 3, Lepetit reflects on failures, celebrates best practices, and offers insights on what it takes to drive sustainable AI success in a global enterprise.

Edited Excerpts

Q: Not many people openly talk about failure. But learning from what doesn’t work is just as critical as sharing success. Could you tell us about a time something didn’t go as planned and what you learned from it?

To be a learning organization, we must embrace failure. We need to fail fast, but more importantly, we need to recognize failure so we can grow stronger from it.

One key challenge we faced was organizational. Early on, we structured teams based on capabilities – engineering, BI, data science. But this made it difficult for internal customers to engage with us. They didn’t know where to go or how to connect their needs to our structure. We realized this was a barrier, and last year we reorganized 150 people into service lines. These lines are now designed around delivering value end-to-end, from access to information to ensuring product quality, so we can better connect with business needs and outcomes.

Another issue was around small to medium-sized projects. These often lacked a clear translation layer, a data product manager who could connect the business ask to technical delivery. We saw quality gaps because of this. That led us to formalize roles, create design authorities, and improve how we ensure professionalism throughout the lifecycle of a data product.

Lastly, we’ve sometimes over-engineered solutions, fine-tuning AI models beyond what was needed. We’ve now put systems in place to avoid wasting time, so we can test, fail, iterate, and create impact faster.

Q: You’ve mentioned the importance of a translator in making AI a reality across the organization. What does that look like at Saint-Gobain?

It starts with listening. I’m passionate about meeting people across the organization to understand what they do and what challenges they face. Some naturally have strong soft skills like curiosity, asking the right questions, and understanding business needs. These individuals often come from BI backgrounds, where they work closely with the business and not necessarily with deep technical expertise.

These soft skills are essential for translation. We’re now investing in identifying and upskilling these talents, helping them grow into roles like data product managers. At the same time, we’re making access to knowledge easier. Generative AI, for example, is a great enabler – chatbots can help bridge the gap between business needs and the right data, technology, or solution.

Q: You’ve also spoken about shifting from optimizing everything to aiming for “good enough.” That’s a big mindset shift, especially for data scientists. How are you making that happen?

It starts with processes. We’re not reinventing the wheel. We’re applying tried-and-tested service delivery principles from traditional IT and BI into the AI space.

From capturing data to exposing it through applications or BI tools, we’ve formalized how we validate whether something meets the business demand. Our data science teams work closely with this customer-facing layer to define stopping points: when does a solution meet the minimum requirements to go to market?

Rather than chasing 96% accuracy over 95%, we focus on iteration, delivering value fast, and then improving based on real-world results.

Q: If you had to pick one best practice from your journey, what would it be?

Always start with the business problem. Too many data and AI initiatives focus on the technology first and fail to translate into impact.

Our approach is to speak the business’s language. We bridge the communication gap with non-technical stakeholders, ensuring they understand what the solution does and how it connects to their goals. Together, we build a value chain, from business needs to execution to financial impact.

And we measure that impact. We show hard savings and quantify how data, IT, and AI projects contribute to the bottom line or grow the top line. That’s how we ensure we’re not just building POCs or MVPs, but driving real business transformation.

Q: Many aspire to become group CDOs or lead data and AI efforts. What personal advice would you give to someone pursuing that path?

Continuous learning is key. You don’t need to know every detail of every AI technology but you must be able to understand trends, connect ideas, and guide your organization through change.

Equally important is the ability to inspire a cultural shift. You need to foster a data-driven mindset across the company and make data accessible and actionable for all employees. That takes curiosity, communication skills, and the ability to connect technical solutions to business challenges.

Stay curious, stay connected, and stay open to learning. That’s what will help you succeed as a CDO, and keep succeeding in a world that’s constantly evolving.

Q: You’re an executive with a full plate. How do you personally find the time to stay curious and keep learning?

I ask myself a simple question: If I’m absent for several days, and nothing changes – if no value is lost – then something is wrong.

That reflection keeps me grounded. It reminds me to keep learning, keep questioning, and stay curious, both internally and externally. Curiosity isn’t a distraction from the job — it is the job because transformation and leadership require us to challenge assumptions and evolve constantly.

CDO Magazine appreciates Benoit Lepetit for sharing his insights with our global community.