Retailers’ AI ambitions stalled by silos, legacy systems—and a culture stuck on pilot mode

Organizational silos and fragmented data infrastructures remain primary barriers preventing retailers and grocers from scaling AI beyond pilot stages. According to the FPT study, 41% of firms cite integration complexity and 38% point to data silos as critical obstacles, while legacy on-premise systems further obstruct the real-time data pipelines essential for AI inference, as noted in Southeast Asia’s enterprise AI challenges. This fragmentation not only stalls AI operationalization but also undermines timely decision-making, with 43% of UK retailers identifying delayed decisions as a key agility limiter during peak seasons, despite widespread AI adoption.

A pervasive lack of strategic alignment and governance frameworks constrains meaningful AI progress within organizations. Only 39% of firms report alignment across strategy, governance, and operating models, while 35% fail to collect quantified AI metrics, impeding informed decisions on project scaling or termination. As FPT’s Pham Minh Tuan emphasizes, overcoming siloed innovation requires integrated governance and operational discipline to transform isolated AI experiments into repeatable, enterprise-wide execution. Without this, many retailers remain trapped in reactive, crisis-driven modes, relying on 'war-room' responses rather than proactive AI-enabled strategies.

Cultural resistance and outdated operating models limit AI’s transformative potential, with many executives viewing AI narrowly as a tool for efficiency rather than a catalyst for redesigning retail workflows. Gary Hawkins of the Center for Advancing Retail & Technology highlights that only 34% of organizations pursue an AI-first operating model, focusing mainly on automation and cost reduction. This cautious adoption is compounded by a disconnect between top-down AI mandates and bottom-up cultural constraints, as Evelina Samuels notes, underscoring that AI governance must address people and culture, not just technology, to truly shift how work and decisions evolve.

Realizing AI’s full value demands deep transformation of business processes and operating models beyond mere technology deployment. Over 80% of senior IT leaders feel confident in AI scalability, yet 75% acknowledge that outdated workflows and enterprise design limit value creation. Experts like Anjali Shaikh and Peter-Paul Schreuder stress that successful AI integration requires mapping workflows, redefining roles, and sustained cultural change—efforts that take months of rigorous process work rather than quick tool training. Companies like IMA Financial Group exemplify this approach by aligning AI adoption with reskilling and work redesign, turning AI from a technical project into a fundamental organizational transformation.

Integration complexity and pervasive data silos remain formidable barriers to scaling AI beyond pilot stages, with 41% and 38% of firms respectively citing these issues as critical challenges. The transition from fragmented AI tools to unified architectures demands coordinated full lifecycle deployment models that encompass strategy, governance, and operations, rather than mere technology acquisition. As the FPT study highlights, nearly half of respondents prioritize providers capable of engineering, deploying, and running AI systems end-to-end, underscoring that governance and operating discipline are as vital as technology access in overcoming integration hurdles.

Legacy infrastructure and siloed databases, especially in regions like Southeast Asia, significantly impede AI deployment by failing to support the real-time data pipelines essential for AI inference. Companies such as SotaTek advocate a phased approach starting with foundational audits of data quality, accessibility, and compliance, followed by targeted deployments on repetitive processes to build trust and demonstrate ROI before broader organizational change. This methodical progression, paired with deliberate investments in API-first architectures and middleware layers, addresses the integration friction between AI models and existing ERP, CRM, or HR systems, enabling seamless AI adoption.

In retail and supply chain contexts, the value of unified AI architectures is vividly illustrated by companies like Everest Truck Brokerage, which leverages Pallet’s AI-powered workflow automation to handle roughly half of its complex multi-step processes. This integration not only counters industry-wide margin compression—from nearly 20% down to 10%—but also creates a triple-win ecosystem where operational efficiency, customer service, and AI provider revenues all improve. Such success stories emphasize that integrating internal operational data with external network signals is crucial for truly intelligent AI decisions, moving beyond isolated data lakes to network-level intelligence.

The future of retail AI hinges on unified data and integrated operations that connect product data, inventory accuracy, order management, fulfillment logic, and customer intelligence into a cohesive platform. Rather than relying on isolated tools, leading retail technology companies are building modular, scalable platforms directly linked to business outcomes, enabling incremental adoption without full system overhauls. As Óscar Hernández of Bluetab LATAM notes, strategic integration supported by cloud scalability and strong data governance frameworks is essential to optimize pricing and product availability while safeguarding profit margins, marking a decisive shift toward AI as the core operating layer of retail.

Back-end operational AI tools are proving to be the real workhorses in retail and grocery, tackling entrenched challenges like shrinkage, labor costs, and fresh product management with clear ROI. For instance, Hy-Vee’s partnership with Relex to enhance fresh product forecasting and Albertsons’ incremental AI adoption—ranging from a proprietary produce inspection tool to an agentic shopping assistant—demonstrate how practical AI applications improve both operational efficiency and shopper experience simultaneously. This pragmatic focus on operational AI contrasts with the flashier front-end chatbots, which, while headline-grabbing, often deliver less tangible business value.

Small and mid-sized retailers can unlock measurable business value by embedding AI directly into specific workflows that address routine but critical operational 'leaks' such as abandoned cart recovery, lead follow-up, and missed calls. Instead of building large AI teams, the strategy is to identify one painful workflow, automate it with AI, measure the results, and iterate—a method that ensures consistent execution and prevents revenue leakage caused by human oversight. As one analysis advises, 'Do not build critical business processes around someone remembering to check a spreadsheet... That is not a process. That is a hope with a calendar invite.'

AI’s most effective retail applications often support decision-making rather than supplant it, exemplified by decision support dashboards that help analysts assess procurement choices, carrier performance, and emerging pricing mechanisms. This approach—highlighted in multiple analyses—enables retailers to pinpoint issues and optimize operations without relinquishing control, fostering greater adoption and ROI. The growing interest in AI-driven pricing agents signals a promising frontier where AI augments human judgment to drive measurable financial outcomes.

Conversational AI and retail media automation are emerging as powerful levers to enhance shopper engagement and operational productivity. Albertsons’ conversational AI search on their website has driven up to 10% basket growth, illustrating how AI can break habitual shopping patterns and inspire new choices. Meanwhile, AI’s automation of repetitive workflows in media buying and analytics frees teams to focus on strategic initiatives, unlocking new efficiencies. Despite some overestimation of AI’s impact on in-store commerce, experts emphasize the underestimated opportunity in transforming insights into actionable decisions that improve customer experience and business outcomes.