Roberto de Angelis is Managing Director & Partner at the Boston Consulting Group in Zurich and a core member of BCG’s global Consumer Goods practice. He advises leading retailers and FMCG companies on growth, portfolio strategy, and innovation across food and non-food categories.
Roberto, what are today’s biggest strategic challenges for retail leaders in Europe?
Retailers in Europe are operating in a highly unfavorable context shaped by three major forces.
First, consumer spending power is eroding. Inflation and stagnant wages are shrinking real income, making shoppers more price-sensitive. They split shopping trips across more stores, buy on promotion, and delay or cancel purchases — leading to smaller baskets, lower margins, and market share loss.
Second, omnichannel expansion is weighing on store productivity. Non-food categories have largely moved online, food retail faces pressure from Amazon Fresh, and many omnichannel models remain complex and unprofitable, viable only in select city locations.
Third, physical store networks are constrained. High capital costs, lengthy renovations, and a saturated market limit expansion opportunities, reducing in-store productivity overall.
To stay relevant, retailers need to act on three imperatives:
- Know your shoppers – Use loyalty data for sharper insights and relevance, create compelling assortments, drive product trials, and strengthen private labels.
- Precision and personalization – Leverage physical proximity to deploy the right offers and assortments locally.
- Efficiency and speed – Shorten decision cycles, accelerate network upgrades, reduce GNFR costs, and build stronger supplier co-investments.
Why Most Retail AI Projects Stall at Pilot Stage
Recent surveys show that, while many (ca. 80%) companies are already using AI, the actual P&L has been elusive, with about 10% of companies agreeing to have unlocked significant value.
Our experience indicates that, as a rule of thumb, the contribution to value creation comes 10% from data, 20% from algorithms and technology, and the remaining 70% from changes in processes and organizations to make changes stick.
Often, too much attention is placed on data and technology, and far too little change is brought about in organizations and process changes. As a result, companies continue to operate with legacy, non-tech-enabled processes, and fail to unlock the remaining 70% of value.
The reason for this is often the lack of central steering, which in turn does not enable leaders to implement the bold changes required to achieve an AI-driven transformation.
Leading companies which unlock significant advantages from AI have a few features in common:
- A CEO-led transformation and a company-wide steering
- A clear vision, roadmap and prioritization of which value areas, business units, functions or processes have potential to be transformed, and how much is the value at stake
- Leverage the nascent AI-ecosystem, to speed up development of technology avoiding costly, internal mega projects that often run late and over budget.
Where Retailers Should Start Their AI Journey
In retail there are already a relatively wide-range of use cases where AI can make an impact, so the starting point really depends on where the biggest value lever lies for each individual company. For some, margin optimization is the obvious entry point, through assortment, pricing or promotions. Others might see the biggest impact first in cost reduction, for example through automation in customer service, planograms or logistics. And for some retailers, customer experience is the priority, with personalization, loyalty and omnichannel journeys. The important thing is to have clear priorities, a roadmap with a value creation “north-star”, and the license to operate while transforming legacy processes which diminish the benefits of AI and automation.
Where European Retailers Can Differentiate
European retailers cannot outspend platforms like Amazon or Temu, but they can differentiate in other ways. One area is the combination of loyalty data with AI to create true personalization for shoppers. Another is building stronger omnichannel journeys that are rooted in local markets and really reflect customer needs. And finally, partnerships with suppliers for closer data-driven collaboration can become an important source of differentiation. In short, European retailers will not win by scale alone, but by speed, adaptation and customer closeness.
Make or Buy: The Right Approach to Retail AI
In a fast-evolving space like AI, “make” is a very risky choice, as the pace of innovation is likely faster than the time to develop models and technology internally. At the same time, owning a unique set of purpose-built systems can drive unique competitive advantage. As a result, and as part of the “roadmap” that each CEO should have, companies should be very intentional on this decision. Prioritize Internal developments if and where it creates real differentiation and if you have the right talent in house. Partnering with the nascent AI-ecosystem enables speed, innovation and cost / capex effectiveness, as well as freeing up resources to focus on process and operating model changes (the “70%”), which requires deep knowledge of how a company operates, and cannot therefore be outsourced.
Spotlight on the Future: AI Winners vs. Latecomers
The retailers who succeed will be those who operate with speed, moving from decisions and implementation in months to cycles of just weeks or days. They will have a tech-enabled operating model that includes automation, robotics and computer vision. And most importantly, they will integrate AI into their actual decision-making processes, rather than leaving it in back-office pilots. Clear leadership from the business side, not just from IT, will also be a defining factor.
One Key Principle for Retail AI Strategy
The most important principle is that companies need to be very clear about where they want to differentiate. Above all, the real impact is unlocked when processes and operating models are upgraded to enable faster, better decisions – that is where the 70% of the value lies, much more than in data or algorithms alone.
Thank you, Roberto!