In Consumer-Packaged Goods (CPG), speed alone is no longer enough. According to Howard Yu, LEGO® Professor of Management and Innovation at IMD Business School in Switzerland, the companies that thrive with AI aren’t the fastest movers, but the most disciplined ones; those that build strong data foundations and make transformation sustainable.

In this interview, he shares his perspective on the attributes that set future-ready CPG players apart, and what it really takes to make the leap into AI.

In 2021, you said: “Going digital is no longer optional.” In 2025, ‘going AI’ seems equally non-negotiable. How can leaders effectively navigate this change?

The world is spinning so fast. During the pandemic, it was all about digital and omnichannel, but now it’s about generative AI and AI interfaces.

What I’ve seen is that the most future-ready businesses aren’t necessarily those moving at the highest velocity. Speed is essential, but the real winners are those who take a step back, think things through, and plan carefully.

It’s the slow work that really matters. That’s today’s challenge: taking a step back and focusing on the critical building blocks. If your organization still runs on manual coordination, with nothing documented, AI can’t help you. You need a proper corpus of data. Some businesses rush into superficial AI projects, partnering with startups to launch Large Language Model (LLM) features, but those solutions rarely deliver a nuanced, compelling customer experience.

The real foundation is capturing the right data, simplifying the organization, and eliminating manual processes. That groundwork enables you to take full advantage of generative AI. Once you’ve built those foundations, LLMs simply accelerate growth.

You’ve spoken about how jobs, processes, and value chains will be decomposed and reassembled like LEGO blocks, with parts automated by AI. Where do you see the earliest and most promising signs of this transformation taking shape? What hurdles must we overcome to scale this reinvention for the 8.2 billion people that the CPG industry serves?

To capitalize on AI’s infinite opportunities, CPG executives need to play a dual game: deliver today, while transforming for the future. Generative AI will profoundly change customer interfaces, workflows, and the workforce. However, if consumers aren’t buying now, you’re in trouble, whether you’re a large corporation, a small-cap company, or a family-owned business.

That’s why AI deployment needs to run on two tracks. First, there’s the short-term track: productivity gains, new features for customers, better support for retailers, and guiding shoppers to the right products. These can be deployed quickly, assuming the groundwork is in place, without requiring heavy senior management oversight, thereby empowering employees to experiment and move quickly

Second, there’s the enterprise-level track: redesigning end-to-end business processes with strategic partners to eliminate steps, reduce manual coordination, and enable junior staff to manage more complex tasks.

If AI deployment is framed purely as cost-cutting, you’ll face resistance. People fear for their jobs. Leaders must instead envision AI to be workforce employee friendly. That means reimagining it as not being about automation, but about empowering people with AI tools so that they can do more meaningful work.

With that value system, the generative AI transformation everyone talks about becomes less painful and ultimately more effective in achieving business goals faster, because we take people along the way.

Most current LLM use cases cluster around customer service, content, and code generation. Can you share any examples of CPG companies applying LLMs in novel, high-impact ways?

In our latest ranking at the Centre for Future Readiness, Procter & Gamble, Coca-Cola, and L’Oréal stood out. They’ve built strong digital backbones, allowing data to flow across silos – and that took years of work.

Procter & Gamble, for instance, spent almost ten years running data visualization initiatives, which built a high level of data literacy among managers. This culture of transparency is a real differentiator. No one has nailed it entirely but, in business, being one inch ahead of the competition is enough.

Coca-Cola built sandboxes worldwide, allowing teams to use their advertising image assets to create locally relevant marketing content, internet memes, and campaign materials to boost conversion rates.

L’Oréal, meanwhile, has long invested in its beauty tech pipeline, such as ModiFace’s augmented reality tools for virtual product trials. Generative AI is now enhancing that functional excellence.

What differentiates these organizations is that they’re not just chasing generative AI, they’re also maintaining expert systems based on traditional machine learning to optimize and clean databases. LLMs are powerful, but not omnipotent.

As we approach a world where AI is universally accessible and free, what will be the enduring sources of competitive advantage for firms?

The fear is this: in a few years, if generative AI becomes extremely smart, it could wipe out direct consumer connections overnight. Imagine someone asks AI to “get me the nicest Christmas gift for my child.” The AI purchases without the customer ever visiting the retailer’s site. The same could happen with airlines or hotels; years of optimized interfaces, marketing campaigns, and customer knowledge could vanish if bookings all happen through AI intermediaries.

That’s the fear. But here’s the key, it’s just an interface change. What really powers LLMs is data. If you’re an operator, there are three ways to think about maintaining advantage:

1. Leverage proprietary data: Use customer data that’s not publicly available so your matches and recommendations outperform generic AI results. For example, if you’re Trip.com, Expedia or Booking.com, you have proprietary hotel reviews, traveler demographics and behavior data. That can produce a ‘wow’ result where a generic AI agent delivers only an average result.

2. Capture new, unique data: With appropriate privacy safeguards, companies can collect unique, real-world data the open internet doesn’t have. Theme parks use wearables to track guest activity. Hotels could gather on-site data to understand what drives satisfaction. Context matters, and cleverness without context is useless.

3. Monetize that advantage: You could keep the data to enhance your own experiences, license it to AI providers or build your own ‘agent within an agent’ so customers seek you out directly – like apps in Apple’s App Store.

The point is: think about data differently. That’s how CPG companies can turn AI into a tailwind rather than a headwind.

For organizations without much technological capability, AI tools let them quickly catch up to the world’s intelligence. What does that mean for large CPG players, now that anyone could compete almost overnight?

Big organizations have a unique opportunity. They need to think beyond just the product, almost in ecosystem terms. If you’re Unilever, you already work with small distributors and mom-and-pop stores. But now you can analyze on their behalf, helping them serve end consumers better.

Every interaction – whether with a customer or a consumer – is a data touchpoint. You need to bottle that exhaust. With generative AI commoditizing open knowledge, the only real advantage is understanding your unique context better than anyone else. If you know how a particular community cooks, eats, or shops, your innovation and recommendations will be more effective. And if you don’t capture that data, you’re invisible to AI.

That means shifting the mindset from product to data and reskilling the entire organization – from top to bottom.

Your research spans industries, regions, and technologies, often identifying emerging patterns well before they become mainstream. I’m curious, looking back over the last five years, what discovery or trend genuinely surprised you?

That future readiness is always a work in progress. If you’re not pushing uphill, competitors can catch up fast. In the pre-internet era, it might take 10–15 years. Now, generative AI accelerates the spread of general knowledge dramatically.

This cuts both ways: if you fall behind, you can catch up quickly if you’re willing to learn. But if you’re ahead and take your eye off the ball, you can lose your lead just as fast.

Proprietary data is what sustains differentiation, your unique, digitized context. And to move fast, your backend must be clean. If your corporate kitchen is messy, full of manual coordination, endless meetings, and undocumented processes, you can’t adapt quickly. Future-ready companies have clear documentation, accessible information, and skilled employees who can act without constant approvals.

What’s the biggest danger for companies that are lagging?

Without hesitation, it’s seeing LLMs as a magic bullet. I’m starting to see that happen. They might decide to buy an LLM company or partner with a technology company, thinking it will save them. They pile their hopes onto one single savior, and that’s risky. We know small-scale experimentation works, but one big bet usually doesn’t.

Going back to your point about ‘keeping the kitchen clean’, what should IT, data and analytics departments do more proactively to support their organization and not feel left behind, while also bringing the organization together and breaking silos?

In the past, the business would come to IT, and IT would provide full services. I still see IT departments doing one-off analytics for senior management. That’s backwards.

Leading organizations like MasterCard, DBS, and L’Oréal are shifting their IT role to developing, safeguarding, and improving new tools, then passing them to the business to deploy themselves. This setup was pioneered at Amazon under Jeff Bezos.

Once upon a time, Amazon’s IT was full-service. By the early 2000s, innovation slowed, and Bezos issued the API mandate: no more spreadsheets flying around, no customized data pipelines. Everything had to run through hardened APIs, or you’d be fired. The result? IT became a self-service infrastructure. Business units could tap in on demand, while IT focused on developing modular, plug-and-play tools.

This is now widespread across future-ready organizations. IT stacks are already modular, tested across markets, and standardized. Business units just plug and play, quality checks are automated. IT then co-develops new tools with the business, stabilizes them, and moves them into the mature stack. It speeds things up and eliminates conflict.

Given that you travel often to Taiwan, China and Hong Kong, What lessons could Western CPG companies learn from how firms in the East are outpacing their global peers?

Two things stand out. First, in the West, generative AI sparks a lot of fear. In China or Singapore, people are excited. In Germany, it’s ‘Stop this monster’. In Shenzhen, it’s ‘Let’s go!’ Why? In Asia, the last wave of globalization brought benefits: outsourcing, manufacturing, and prosperity. People are used to creative destruction. As long as you learn, you can move up the value chain.

In the West, historically, they were the industrialized leaders. That’s where today’s political debates come from. So, when people look at AI, reactions differ.

Inside a company, if you face employee resistance, you need to frame AI adoption gently. The Western narrative, especially in Silicon Valley, jumps straight to artificial general intelligence (AGI) wiping out humanity. That’s nonsense. AGI isn’t ready. There’s enormous value in expert systems that make work more enjoyable.

Science may be neutral; technology isn’t. Excessive automation that eliminates jobs without better prospects is evil technology. Companies can choose differently – focus on tools that augment creativity and skills.

Second, in Asia, AI adoption doesn’t have to be highbrow. Even basic pattern recognition and data labelling are fine if there’s a business case. In Europe, there’s often hesitation until the ‘perfect’ solution appears – which leads to inaction. Mindset makes all the difference.

You’ve worked closely with global business leaders and tracked their ability to future-proof organizations. From your vantage point, what is the one question you wish more CEOs or executives would seriously reflect on, but tend to overlook?

CEOs need to think end-to-end about transformational AI business processes. Without CEO or top-team intervention, the architecture of a large organization will always default to silos. Business managers will deploy technology within their silo. The CEO must push for cross-silo transformation.

A litmus test: Look at your cross-silo AI workstreams. If the owner is just ‘an IT guy’, I’m not convinced. If the owner is the chief revenue officer, chief marketing officer, or a regional president, someone close to the customer, then I am. That’s rare, because many business leaders don’t want to learn the tech, especially if they’re close to retirement.

Future readiness can be measured. If your five must-win AI battles are all owned by IT, you’re not ready. The business needs to own them, build the right business cases, and package proprietary technologies for the AI world.

‘LEAP: How to Thrive in a World Where Everything Can Be Copied’ is available now from all good bookstores. For more insights, Howard Yu shares his views frequently on his substack called: One Inch Ahead

https://howardyu.substack.com/

You can also read Howard’s Future Readiness Report for CPG organizations here: https://www.imd.org/future-readiness-indicator/home/consumer-packaged-goods-2025/