Published on the 10/12/2025 | Written by Heather Wright
Frameworks needed to turn pilots into measurable gains…
Retailers are investing heavily in AI, but a new report warns adoption doesn’t always equal impact and urges executives to remain ‘hyper-focused’ in order to realise the benefits of the technology.
The Berkeley Research Group report says while nearly half of retailers have deployed AI across core functions, any lack the operating model, governance and measurement frameworks needed to turn pilots into profit.
“Without discipline, AI risks becoming a series of disconnected experiments rather than a driver of enterprise value.”
The report, which surveyed 100 executives across North America, Asia-Pacific and Europe, found North American retailers are currently using AI for marketing (70 percent), IT/digital (62 percent), digital commerce (56 percent) and merchandising strategy and pricing (54 percent). When it comes to how they plan to use it, however, it’s a different story, with planning and product flow the leading desire at 40 percent. That’s followed by corporate functions (38 percent), supply chain and sourcing (36 percent) and distribution and logistics (32 percent).
The report notes APAC retailers – and those in Europe – trail their North American counterparts in use, with APAC respondents also the least bullish overall about AI’s long-term industry benefits. The most common implementation area, and the largest share of any regional group, was IT and digital (68 percent), likely driven by persistent cybersecurity threats. APAC respondents had the lowest level of AI implementation across stores and services (32 percent), planning and product flow (28 percent) and corporate functions (20 percent).
That’s in keeping with a UKG report late last year which found just 33 percent of Kiwi retail workers, and 46 percent of those in Australia, were using AI and automation in the workplace – well below Singapore’s 92 percent. That survey, which included 213 Kiwi retail workers and 252 Australian workers, saw Australian users citing cited as the biggest benefit, improving inventory management, scheduling and time, attendance, leave and accruals. Among Kiwi respondents, there was a clear desire to use AI to improve staff management processes followed by customer sentiment analysis and automating inventory management.
The bosses, too, are keen: Salesforce’s Connected Shoppers survey earlier this year saw 77 percent of A/NZ retailers flagging AI agents as being ‘essential’ to compete within a year, and 74 percent say they’re increasing AI spending this year accordingly.
However, that spend, and even increased AI usage, doesn’t necessarily translate into tangible business impacts, according to BRG.
“The widespread use of ChatGPT and Copilot for relatively mundane tasks like drafting product descriptions, purchase orders or marketing copy, my offer some value. But is it a foundational shift?”
Measuring impact
That question – whether AI is delivering a foundational shift or just incremental efficiencies – is a question for all organisations, and it sits at the heart of BRG’s research. The report makes it clear that adoption is widespread, but impact is uneven and many retailers lack the operating model, governance and measurement frameworks needed to translate experimentation into enterprise value.
The report cites several key performance indicators it says retailers should consider using to measure AI’s benefits, including average order, customer retention/repeat purchase rate, inventory turnover rate, forecast accuracy and gross margin improvement.
“Companies should fit AI into a clearly defined target operating model, rather than force the technology into what they’re already doing,” BRG says. “AI is not cheap, and payback periods vary drastically; retailers must have a clear roadmap with well-defined business cases, ROIs and pilots to test and launch their AI initiatives.”
As to the key AI opportunities for retail, BRG says process flow automation, to reduce data entry and off-systems data manipulations to feed the next systems is a significant opportunity for AI enhancement, given is a significant drain on many retailers.
“Retailers must remain hyper-focused on embedding AI into operating models, strengthening governance and defining new KPIs,” BRG says. “Without that discipline, AI risks becoming a series of disconnected experiments rather than a driver of enterprise value.”
Using predictive AI to leverage real-time store- and channel-level data along with external data to provide more dynamic demand planning and nimble execution of allocations, markdowns and promotions; and providing enhanced customer engagement via seamless omni-channel experience such as virtual try-ons and personalised services and to establish reliable product reviews/feedback is also noted.
Enhanced supplier management and supply chain oversight, enhanced design processes eliminating the need for manual sketches and improved pricing and promotion strategies are also key opportunities.
But BRG also warns that poorly implemented AI can create more confusion than clarity. Retailers can use AI to act more precisely on promotions that drive real results, but companies often bring in applications and run them in siloes with existing merchandisers using traditional processes ‘and in the end, no one knows what the prices should really be’, it says as an example.
The report highlights some retail success stories – predominantly among larger retailers – including membership-only warehouse club retail chain Sam’s Club (a division of Walmart) which has ‘reinvented’ checkout lanes across 600 stores using AI-enhanced scan and go app validation, and Levi Strauss, which is using generative AI to produce first drafts of product descriptions and translations. German grocery chain REWE has used AI-powered demand forecasting to reduce the number of unavailable items by half and is adopting AI at self-checkouts to prevent theft, while AI driven personalisation tools have seen beauty retailer Sephora increase average order values by 25 percent and customer satisfaction by 20 percent.
A/NZ stories
Locally, companies such as Coles, Wesfarmers and The Warehouse Group have been adopting AI.
Coles has a suite of AI models driving operations, including processing 1.6 billion predictions daily across 20,000 SKUs and 850 stores using AI to optimise replenishment and service, and the use of computer vision for produce recognition and queue monitoring.
Competitor Woolworths Group signed a new deal with Google earlier this year to deepen data-driven insights and automation at the retailer.
Wesfarmers meanwhile reported a number of active use cases across its portfolio late last year, from Bunnings’ internal ‘Ask Lionel’ instore assistant, which provides staff with access to product data, insights and announcements to answer complex customer queries, to multilingual live chat, personalised offers and conversational commerce capabilities.
The company did however find itself in hot water recently when the Bunnings AI assistant provided electrical advice meant only for qualified professionals to a customer.
In New Zealand, the Warehouse Group has touted its AI-enabled BI/predictive analytics, use of genAI for product copy and faster insight cycles. It’s currently upskilling its workforce as part of a deal with Tata Consultancy Services earlier this year.
As BRG’s report notes, while AI is reshaping retail locally and globally, its success will hinge on disciplined execution – something that applies not just in retail, but across all business.
