AI in Retail: Where to Invest for Operational Impact

Retail leaders are no longer asking whether AI matters for retail strategy. The real question is how retailers should prioritise AI initiatives within a broader AI strategy, and how to ensure those initiatives deliver measurable value rather than noise.

ADAPTOVATE spoke with Marcella Larsen, CEO of The Woven Group; Matthew Harris, Retail and CPG Senior Account Executive at Microsoft; Mandy Flatley, retail executive and former People Director at Woolworths; and Connor Hershkowitz, Lead Consultant at ADAPTOVATE, about how retailers are aligning AI strategy to operational impact. The conversation moved well beyond theory, focusing on capability gaps, governance, frontline implications and the structural shifts now underway in the C-suite.

Start with outcomes, not tools

Connor Hershkowitz pointed to a recent engagement with a national retailer employing 16,000 people across four brands. The work centred on the people and safety function, which supports both head office and frontline teams and oversees safety across tens of thousands of products.

The symptoms were familiar. Teams were “focused on delivering a high volume of outputs rather than the outcomes” that mattered most. Executive-level changes were driving shifting priorities. Capacity was limited. At the same time, there was a strong internal narrative that AI would fix the problem.

“There was a sentiment within this organisation of ‘AI’s here, let’s just use AI, it will fix it,” Hershkowitz said.

Instead of starting with technology, the ADAPTOVATE team focused on defining AI use cases aligned to strategic business outcomes, supported by quarterly delivery cycles anchored in objectives and key results.

Within two quarterly cycles, delivery of strategic objectives improved from under 25 percent to more than 80 percent. Three were prioritised. One involved reviewing safety and compliance data across 60,000 products, a task previously managed manually through spreadsheets. Another addressed HR policy queries across 16,000 employees, reducing email enquiries by 60 percent in one month after introducing an automated, AI-supported solution. A third outcome was cultural: the people function positioned itself as a role model for embedding AI adoption into the retail operating model, rather than treating it as a standalone tool.

Critically, the retailer stopped 11 lower-value initiatives to free up focus and resources. AI was given strategic attention rather than bolted on to an already overloaded agenda.

Understanding the types of AI

Marcella Larsen emphasised that much of the confusion in the market stems from treating all AI as the same.

Many retailers have been using analytical AI within retail pricing and forecasting systems for years. That form of AI is structured, predictable and grounded in rows and numbers. Generative AI, however, operates on large language models trained on unstructured data such as documents, images and text. It creates content and supports productivity but introduces challenges around reliability and governance.

“These are trained on large language models. There are hallucinations,” Larsen said. As organisations scale generative AI adoption across the enterprise, data discipline becomes even more important. The emerging frontier is agentic AI, which automates sequences of tasks and moves closer to outcome-driven autonomy. Larsen cautioned against overestimating its maturity. “Truly autonomous agentic is when it will fully have AI oversight,” she said. “I think we’re a couple years out on that”. While the vision is compelling, most current implementations still rely on a human in the loop.

Why retail urgency has accelerated

For Mandy Flatley, the acceleration in AI adoption in retail is driven by competitive pressure and rising customer expectations. Whoever can remove friction for customers and teams will win.

The landscape is also shifting in how consumers discover and buy products. Matthew Harris described an example shared at a recent industry forum: a consumer searching for headphones using a traditional search engine received one set of results, influenced heavily by advertising spend. Using an AI-powered shopping agent, the same query produced different brand recommendations. When asked why a well-known brand was not included, the agent replied that it “might be the best at spending money with Google, but they’re not necessarily the best headphone provider.”

The implication is significant. If AI-powered shopping agents begin to dominate product discovery, brand visibility will depend less on paid search. and more on product quality, reviews and trusted signals across the web.

Larsen pointed out that Australia currently has one of the highest consumer adoption rates of AI-supported search globally. Discovery and purchase are increasingly converging. “They’re fully prepared to buy,” she said, describing customers who arrive via AI channels as already qualified. This shift forces retailers and consumer goods companies to rethink website architecture, data structure and how they build trust beyond their own platforms.

The C-suite is being reshaped

As AI transformation moves from experimentation to enterprise priority, executive structures are beginning to change. Flatley observed that traditional vertical silos are breaking down. Retailers historically organised around functions, but value is created horizontally across the customer journey.

She noted that in some markets, chief people officer and operations roles are merging, reflecting the reality that leading people and running process are deeply intertwined. In other cases, chief people officer and chief information officer roles are converging, acknowledging that technology and human capability can no longer be managed separately.

Hershkowitz added that many executives did not reach the C-suite because of their AI expertise. They built careers in marketing, operations or people leadership. Now they are expected to lead on a fast-moving technical agenda. Part of the work involves upskilling senior leaders on what AI can and cannot do, while simultaneously grounding initiatives in business outcomes.

Larsen described the challenge in three stages: framing, identity and evaluation. Leaders must clearly define what AI means for their organisation, align it to their values, and determine how it will be measured. Measurement may differ between the board level, which focuses on return on investment, and departments, which track adoption, experimentation and incremental improvements.

Data remains the non-negotiable foundation

Across the discussion, one theme surfaced repeatedly: data strategy is inseparable from AI strategy.

“There is no AI strategy without a data strategy,” Larsen said. No matter how advanced the vendor or platform, organisations that lack clean, trusted, accessible data will struggle to extract value.

This applies equally to customer-facing AI use cases in retail and internal productivity tools. Retailers exploring predictive forecasting, digital twins of supply chains or automated store operations require robust data foundations to simulate, optimise and act with confidence.

Human skills as a competitive advantage

While much of the conversation centred on systems and structure, Flatley argued that AI will heighten the importance of human capability rather than diminish it.

Frontline teams often face friction from clunky systems or poor inventory visibility, which distract them from serving customers. AI-enabled retail tools can remove some of that friction. But what will differentiate retailers, she said, is not just technical competence but human skill.

“It is going to be the thing that makes the difference,” Flatley said, referring to empathy, listening and collaboration. She pointed to global research showing that so-called soft skills are becoming critical as automation increases. In retail, where experience and trust drive loyalty, the blend of art and science remains central.

Harris agreed that trust will be fundamental as consumers rely more on AI agents to make recommendations. If agents lack context or intelligence, confidence erodes. Retailers must ensure that the information powering these systems is accurate and relevant.

Where to begin

For organisations still at the starting line, the panel’s guidance was pragmatic. Evaluate AI initiatives and AI use cases through three lenses: is the initiative viable in terms of return, is it feasible to build and support, and will it be embraced by employees and customers? Without change management and cultural readiness, even technically sound projects will stall.

How Retail Leaders Should Prioritise AI Initiatives

As AI transformation moves from experimentation to enterprise priority, retail leaders face a practical challenge: deciding which AI initiatives deserve investment and which should wait. Many organisations begin with enthusiasm but quickly find themselves running dozens of disconnected pilots with limited operational impact.

A more effective approach is to prioritise AI initiatives and AI use cases through three lenses: viability, feasibility, and adoption.

Viability asks whether the initiative will generate measurable value. Leaders should assess whether an AI use case meaningfully improves customer experience, operational efficiency, or revenue growth. In retail environments where margins are tight, initiatives that deliver clear ROI tend to gain momentum quickly.

Feasibility focuses on whether the organisation has the data, systems, and capabilities required to implement the solution. Many promising AI ideas stall because data is fragmented or governance structures are unclear. Retailers that invest early in strong data foundations are better positioned to scale AI initiatives across merchandising, supply chain, and store operations.

Adoption evaluates whether employees and customers will use the solution. Even technically sound AI systems can fail if they introduce friction into daily workflows or lack trust from frontline teams. Change management, leadership alignment, and clear communication play a critical role in turning promising AI experiments into sustainable operational improvements.

Retailers that apply this prioritisation framework can move beyond scattered experimentation and focus on AI initiatives that deliver measurable operational results.

Retail will continue to blend art and science, but the retailers that integrate technology with clear intent and human capability will be best placed to convert AI ambition into operational results.