Executive summary
Every global business wants to scale AI. But few are prepared for the part that makes it work: data. In this episode of In other words, Willem Koenders, Global Leader in Data Strategy at ZS Associates, joins host Jason Hemingway to explore what it takes to build data foundations that support growth across diverse markets.
Across more than 100 organizations, Willem has seen the same pattern repeat. Companies underestimate the importance of data governance, delay the business case, and assume global frameworks will apply everywhere. He argues that sustainable AI and global expansion require “data governance by design,” making governance a core part of how teams scale responsibly.
From balancing global consistency with local realities, to embedding data ownership into operating models, to using AI to accelerate the unglamorous work of classification and metadata, Willem offers a clear set of steps for leaders navigating transformation.
His message is direct. AI is only as good as the data behind it.. And the companies that win will be the ones that fix the data first.
The conditions for global AI success
Every company wants AI to drive growth. But as Willem explains, many overlook the prerequisite of clean, trusted, well-governed data.
“You can have the most advanced models in the world,” he says, “but if your data isn’t ready, none of it will scale.”
His early career in corporate strategy opened his eyes to how much transformation depended on data long before AI took center stage. Working alongside some of the first Chief Data Officers more than a decade ago, he helped define what their roles should be, and how organizations could navigate data complexity that stretched across functions, regions, and regulatory environments.
What struck him then remains true now, data sits at the intersection of technology and business, and most strategic objectives simply cannot be met without addressing it.
The biggest governance mistake leaders still make
Across continents and industries, Willem sees one misstep more than any other, delaying the business case.
“Data governance still has the connotation of red tape. Push the business case into year two and the next funding cycle dismantles half the work.”
His advice is unequivocal:
- Articulate the value early
- Quantify the outcomes
- Link governance directly to revenue, customer relevance, and AI capability
Equally problematic is a one-size-fits-all approach to global scaling.
“You need minimum consistency, but also local adaptation,” Willem explains. “Governance never works as a big central team doing the work for everyone else.”
Companies that try to standardize everything from headquarters quickly find that their frameworks collapse under the regional regulatory, cultural, and operational realities.
Governance by design: the framework that works
To change how leaders see governance, Willem uses a simple but effective model of
offensive and defensive capabilities.
Defensive goals include managing sensitive data, complying with privacy laws, and controlling access.
Offensive goals include improving customer experience, enabling AI, and supporting new product design.
Both depend on the same foundations:
- Knowing what data exists
- Understanding where it sits
- Defining it consistently
- Determining when and how it should be used
“When leaders see governance as the enabler of both protection and innovation,” Willem notes, “that’s when adoption finally happens.”
Why AI depends on doing the unglamorous work
Simone contributes an important observation from the product and AI side. LLMs can generate convincing outputs from generic internet data, but meaningful enterprise value comes from clean, proprietary data.
“If your data is noisy or duplicated, the model will produce bad outputs,” she says.
Her point reinforces Willem’s central thesis that data quality, governance frameworks, metadata discipline, and curation unlock the commercial value of AI, not the model itself.
That’s why Willem emphasizes data governance by design, not retrofitting it at the end of an AI programme.
“You cannot start now,” Simone adds. “You should have started two years ago.”
Scaling globally while respecting the local
Willem has lived and worked in Europe, the United States, Latin America, and North Africa, experience that has shaped his understanding of how differently data behaves across regions.
- Regulation varies.
- Risk appetite varies.
- Customer expectations vary.
He argues that global data strategies must be translated into local models of ownership and accountability.
His rule of thumb is simple. “If you create it or change it, you own it.”
Simone adds that sophisticated organizations treat markets as risk tiers, adjusting data and AI strategies based on the value and sensitivity of each market.
Together, their perspectives highlight the need for a tiered operating model which includes one global foundation, multiple local expressions.
AI that complements the human, not replaces it
Willem believes AI should support the people doing the work, not add complexity or override expertise.
For Willem, successful AI integration requires mapping real processes, respecting existing expertise, choosing tools that match local workflows, training teams continuously, and perhaps most importantly, finding practical, high-impact entry points
CRM automation is one of his favourite examples.
“These systems have been terrible for adoption for decades,” he says. “Now with language models, people can update entries directly from a Teams chat. That’s transformational.”
Ethics, trust, and customer confidence
Willem sees two essentials for responsible, trust-building data use:
1. Transparency
Explain clearly what data is collected, why it’s used, and how it adds value.
“Use that moment as an opportunity to engage the customer,” he says.
2. Articulating value
Show customers that giving data results in better outcomes, whether relevance, speed, or experience.
Regional expectations differ greatly, and Willem believes it’s a leader’s job to adapt, not assume.
Simone emphasised the importance of transparent pipelines as customers increasingly want to know how AI decisions are made.
Leading across cultures
Willem’s international career has shaped his leadership ethos.
In the U.S., you need to get to the point quickly.
In Latin America, trust builds slowly.
In some regions, language shapes the work itself, as Willem discovered when he realized Spanish has no distinction between “responsible” and “accountable,” complicating governance conversations.
In short, it’s important to listen first, then adapt.
Looking ahead: the future of data and AI
When Willem looks forward, he sees the next wave of AI value coming not from bigger models, but from AI applied to data itself.
Classification, metadata, lineage, and definitions (once the most manual parts of data management) can now be automated in seconds at scale. And often with better quality than before.
“It’s transformational for building things like a true Customer 360,” he says
Simone added that the future will also favour smaller, specialist models, not general-purpose ones, anchored in organizations’ clean, proprietary datasets.
Planning growth before deployment
For Willem Koenders, the path to global, AI-enabled growth starts long before a model is deployed. It begins with data foundations. Governance embedded from day one, local realities incorporated into global frameworks, and a shared organisational understanding that data is a strategic asset, not an IT problem.
Or, as he puts it:
“AI doesn’t work without the data. Be deliberate about it, and treat it like the business asset it is.”
Hear more
If you’ve enjoyed this look at the highlights from the latest episode of In other words, you can hear the entire episode now on our website, or subscribe via Spotify, Apple, or wherever you listen to your favorite podcasts.
