Retail & AI: Accelerating the scale-up

AI’s impact on retail is still unfolding. According to a recent Arthur D. Little (ADL) survey of Procos members,[1] while nearly four out of 10 have launched concrete applications, the majority remain in an exploratory phase. Yet, pioneers that have established the right human, data, and organizational foundations are already reaping tangible benefits across the entire value chain. In this Viewpoint, we examine use cases, barriers, and success strategies for using AI in the retail sector.

Generative AI (GenAI) has brought retail into a phase of profound and tangible transformation. At the National Retail Federation’s annual conference, NRF 2025, nearly 30% of the solutions presented integrated AI. These included demand forecasting, logistics, dynamic pricing, marketing automation, enhanced customer experience, and others. AI is no longer a technological mirage, but a strategic lever whose applications span the entire retail value chain and are already being deployed by industry leaders.

However, an ADL survey of Procos member retailers between June 2025 and September 2025 reveals uneven adoption.[2] While 39% of respondents already have operational use cases, 61% remain in the exploratory or proof-of-concept (PoC) phase (see Figure 1). Therefore, despite tangible evidence of value creation, the use of AI in retail is still in its infancy.

Executives should view these findings as a warning: waiting to explore AI’s potential in the retail sector carries a high opportunity cost, as pioneers are already consolidating a competitive advantage. The challenge for CEOs is twofold: they must rapidly engage in the AI shift to avoid falling behind, and they need to structure their AI roadmap to turn experimentation into sustainable results.

Therefore, we must approach AI as we would train for a long-distance race: with discipline, careful sequencing, and steady progress toward the goal.

USE CASES ACROSS THE VALUE CHAIN

Concrete & diverse applications

The study confirmed the presence of AI across the entire retail value chain. The main use cases identified by retailers fall into five major domains:

1. Operations and supply chain. Retailers prioritize improving stock management, optimizing logistics, and detecting anomalies more rapidly. Measurable gains include improved forecast accuracy of between 10%-20% and improved on-shelf availability by 10%-15%, contributing directly to sales growth and optimizing working capital.
2. Marketing and communications. Large retailers use AI primarily to refine targeting and segmentation, while smaller players focus on generating and adapting content. In both cases, the goal is to boost marketing performance by reducing costs while increasing campaign effectiveness. In contrast, assortment planning supported by AI has not yet emerged as a top priority.
3. Physical and digital sales. Retailers use AI to better anticipate sales volumes and provide personalized recommendations. These levers aim to stimulate net sales, increase basket size, and optimize margins. However, traffic generation in physical stores, heavily impacted by evolving consumption patterns in recent years, has not yet surfaced as a priority use case, lagging more immediate goals of conversion and margin optimization.
4. Customer service and after-sales. AI automates aspects of customer interactions and facilitates the analysis of recurring dissatisfaction drivers, enabling retailers to enhance customer experience, improve satisfaction, and reduce operational costs related to after-sales service.
5. Support functions. Finance, administration, and HR departments leverage AI to automate repetitive tasks, strengthen the reliability of analyses (e.g., forecasting and error detection), and provide internal chatbots to offer daily support.

High satisfaction for pioneers that scaled

This diversity of use cases shows AI’s immense potential but also uncovers a useful pattern among respondents: value only materializes when initiatives move beyond the PoC stage. Projects that remain pilots will have limited impact. Once deployed at scale, however, they deliver tangible results.

Overall, 55% of respondents reported satisfaction with their AI initiatives; this figure rises to 72% among more advanced players and to 84% for operational use cases already industrialized (see Figure 2). Satisfaction is high across all functions, with particularly strong results for operations and supply chain applications. Thus, pioneers demonstrate that the benefits are concrete: improved accuracy, operational efficiency, better shelf availability, optimized working capital, and enhanced customer experience — all translating directly into revenue growth and significant cost savings.

Efficiency gains: The investment priority

Three main drivers of investment — operational efficiency, decision support, and sales enablement — dominate these five use cases. Priorities vary depending on the size and technical maturity of the retailer, but the underlying logic remains the same: start with visible quick wins to demonstrate value and build momentum, then progressively expand the portfolio to more complex and transformative use cases.

To achieve these goals, however, retailers will need to increase their modest AI budgets. Barely one in 10 companies surveyed invest more than €250,000 per year. Even among intermediate and large groups (turnover >€500 million), six out of 10 spend less than €250,000 annually (excluding core data and infrastructure investment). These figures reflect a clear under-allocation of resources, which limits scaling and slows transformation.

BARRIERS TO DEPLOYMENT

While AI’s benefits are real, adoption by retailers still faces multiple obstacles. Our study highlights four main categories of barriers:

1. Human. AI adoption has been significantly slowed by lack of awareness of use cases and insufficient internal skills. Employees often do not know how to integrate AI into their business practices or how to measure its concrete impact.
2. Data. Concerns about confidentiality and data quality remain a major barrier. AI can only create value if the data is reliable, secure, and compliant with regulatory requirements, especially the EU’s General Data Protection Regulation (GDPR).
3. Operational. Retailers cite a lack of time to identify and assess available solutions, perceived high costs, and the complexity of technological and partnership choices.
4. Cultural. Resistance to change continues to be a fundamental limiting factor. Companies struggle to mobilize teams and to instill lasting trust in AI-driven outcomes.

These barriers differ depending on company size. Smaller retailers highlight costs and a lack of resources, while larger players emphasize insufficient internal skills. For executives, the conclusion is clear: without a structured plan to overcome the above obstacles, AI will remain confined to pilots and fail to deliver impact at scale. AI initiatives must therefore be accompanied by a transformation agenda combining technology, organization, and change management.

A TAILORED PATH TO RETAIL SUCCESS

Prioritize before testing

Ensuring the success of AI adoption first requires a structured approach to prioritization and deployment, balancing what can be developed internally or outsourced and with which partners. Two key steps must be activated:

1. Identify opportunities through active monitoring of solutions used by competitors and in other sectors. An internal survey among teams can map existing use cases and pain points to automate or optimize, and reviewing the strategic plan can ensure alignment with the company’s commercial and organizational priorities.
2. Select and prioritize use cases by defining selection criteria (e.g., expected benefits, feasibility, costs, risks, and likelihood of competitor adoption), mapping use cases by impact and urgency, and finally prioritizing five to 10 key cases that detail organizational implications and deployment approaches (internal, external, custom).

Sequence projects & leverage shared use cases

This approach anchors transformation into a logical sequence: create value quickly with simple use cases, demonstrate impact, then prepare the organization to roll out more sophisticated possibilities. For example, recent client projects implementing AI in after-sales services demonstrate that value is built by starting with quick wins (e.g., intelligent interactive voice response, AI-powered chatbots, or simple task automation) before orchestrating more ambitious initiatives such as augmented agents or advanced customer journey personalization.

This sequencing must also rely on the logic of use case clusters (synergies between data, systems, large language models, and partners), designed to identify interdependent sets of use cases in order to pool investments and avoid fragmented efforts.

Go beyond use case selection

Making thoughtful choices about use cases is not enough. Success also rests on four critical enablers (see Figure 3):

1. Secure the data. AI’s success depends on using reliable data, which must be accurate, complete, consistent, compliant with regulations, and supported by robust underlying systems and processes.
2. Engage the teams. Implementation cannot succeed without real buy-in from employees, which requires tailored training, communication, and change management initiatives.
3. Structure governance and execution. Transformation must be structured to organize the progression from PoC to full-scale deployment into successive stages, with clear exit criteria and systematic measurement of KPIs and outcomes.
4. Anticipate obligations. To embed adoption within a sustainable framework of trust, companies must anticipate regulatory constraints (notably GDPR), ensure algorithm transparency, and address environmental, social, and governance (ESG) considerations (particularly because AI is highly energy-intensive).

Conclusion

ACCELERATE OR FALL BEHIND

AI in retail has reached a tipping point: while early adopters reap measurable benefits, many retailers remain in testing phases.

The key challenge is no longer whether to act but how quickly and effectively. Success hinges on prioritizing use cases aligned with business needs and maturity, securing robust data and organizational foundations, and tracking impact rigorously. AI adoption is less about technology and more about strategic and organizational transformation. Those slow to act risk falling behind as AI becomes a structural driver of performance in retail. Vital imperatives for effective AI adoption include:

1. Prioritize use cases aligned with strategic needs and maturity of your organization.
2. Build strong data and organizational foundations.
3. Measure impact of “your successes” rigorously before going at scale.
4. Treat AI as a strategic transformation, not just a tech deployment.
5. And act quickly!

By David Benichou, Edouard Moulle, Samy Katz, Aurélien Moralis

Notes

[1] Procos is a French trade federation representing specialized retail chains, providing industry insights, advocacy, and support for retail development across France.

[2] The survey consisted of 54 respondents drawn from Procos members, a French federation representing specialized retail, which brings together 310 banners with a combined turnover of €110 billion.

• By Arthur D. Little
• 10/12/2025
• Retail & AI