Stop Designing Store Tech for Customers First.
In today's issue:
The 3-wave framework to bring AI in retail
From “onboarding chaos” to AI-powered training across 100+ stores in 4 months
How a mid-sized retailer turned siloed data into €100K monthly revenue lift
1. The 3-wave framework to bring AI in retail
RETHINK Retail, Session: AI in Retail - The Human Touch (Nov. 11, 2025)
For: C-suite executives, Digital transformation leaders, CIOs/CTOs, Retail strategy teams, Store operations VPs, Supply chain executives
TLDR:
Remzi Ural, Partner at PwC laid out a clear framework for how AI will transform retail: Wave 1 (efficiency and productivity), Wave 2 (better outputs, not just faster processes), and Wave 3 (fully autonomous AI agents running the business).
Most retailers think they're preparing for Wave 1. They're wrong – Wave 1 is already here.
Meanwhile, Tesco reduced a 25-person product returns team to 2 people while cutting processing time from 6 months to 15 days. Service Now reports that 55% of retailers using agentic AI improved gross margins versus 22% without it.
Here's the framework many retailers miss:
Wave 1 is here – and it's already delivering ROI in months. Sumit Mitra, CEO at Tesco Business Solutions shared a perfect Wave 1 example: product returns processing. Tesco receives ~300,000 product returns across stores requiring manual categorization by supplier. The process took 6-7 months with 25 people.
Using AI with vector modeling to identify patterns like "cloudy in milk" or "yellow," they automated the entire process. The result? Two people manage it now (instead of 25), and cycle time dropped from 6 months to 15 days. But those 2 remaining people aren't doing data entry anymore – "they're talking to suppliers about trends, working with buyers to improve products." AI eliminated manual work and elevated humans to strategy.
At American Eagle, Uttam Kumar described predictive restocking: "Data shows that one shelf is about to be empty in 1 hour. AI predicts sales loss and sends an alert to an employee: 'Go restock shelf X.'" That's preventing problems before they occur.
Wave 2 is next – "better, not faster" requires reimagining processes. PwC's Remzi Ural explained that Wave 2 is "not generating faster processes, but better output." Most retailers are still in a Wave 1 mindset: automate what we already do to do it faster.
Wave 2 asks: What becomes possible if AI handles all structured work and humans focus entirely on judgment and creativity? Tesco is building a "fashion AI model" by collecting data from customers' digital wardrobes – what they own, wear, and why. If successful, that dataset can inform product development and inventory placement at granular levels: "Not just style in Germany versus UK, but Stuttgart versus Munich." That's not faster merchandising – it's fundamentally better merchandising.
Wave 3 will require new org structures → prepare now by shifting to activity-based work. Ural's prediction: "By 2028, agentic AI will likely handle 15% of day-to-day work decisions." Wave 3 is when "autonomous AI agents drive the business."
When you reach that state, organizational design changes completely. "Work will be broken down into smaller modular units – discrete pieces of value creation that can be solved by AI, humans, or a combination." Emerging roles like "client solution generalists" reflect activity-based work rather than functional silos.
Mitra's recommendation: "AI is 30% tech and 70% culture. Not everybody needs to code. Some need to understand what AI does for them. Others need to implement AI and manage change."
What to do about this:
→ Audit your current AI initiatives against the three-wave framework. Map every AI project to Wave 1, 2, or 3. Target a portfolio: 70% Wave 1, 25% Wave 2, 5% Wave 3 experiments.
→ Identify your highest-volume manual processes and reimagine them end-to-end. Use Tesco's example: don't just automate parts of returns processing – redesign the entire workflow.
→ Build a proprietary dataset that AI needs but doesn't have. Like Tesco's fashion AI model, identify unique datasets you can create that would enable Wave 2 capabilities.
→ Start shifting language from "roles" to "activities" in job descriptions. Pilot activity-based job architecture: Which tasks are rule-based (candidate for AI)? Which require judgment (human)?
2. From “onboarding chaos” to AI-powered training across 100+ stores in 4 months
Growy App, Session: AI on the ground - Retail (Nov. 11, 2025)
For: Multi-unit retail operators, HR/Training leaders, Franchise operators, Convenience/QSR executives, Retail operations VPs
Background: David Tabone, CEO at The Convenience Shop Group, runs 100+ convenience stores across Malta with over 1,000 employees. Four months ago, his HR team spent every day in literal queues – interviews, onboarding, contract signing, training – in multiple languages, for hundreds of new hires as they scale toward 150 locations. Today, they've centralized all policies and training into an AI-powered platform that delivers consistent content in any language, tracks every employee's progress automatically, and frees the HR team to focus on strategic work instead of repetitive explanations. The implementation took 4 months.
TLDR:
Start with stakeholder buy-in, not technology – Tabone insisted on full alignment from board and senior management before selecting any tools, because "you cannot approach a project like this unless you have full alignment."
AI solves standardization and language barriers simultaneously – content delivered in English, Italian, French, German, Hindi ensures every employee understands messaging perfectly while centralized knowledge bases guarantee consistency.
Pilot with small groups but commit to full rollout quickly – they tested with stores that excelled at learning and stores that struggled, gathered feedback, tweaked, then scaled company-wide.
Here's the framework many retailers miss:
Sell the vision first, choose the technology second. When Tabone joined My Convenience in January 2024, he saw queues of people every day for interviews, onboarding, training. "My team spending their days talking to people, explaining in different languages, dedicating time to less productive stuff."
But most transformations fail because they start with technology. Tabone started with the Board and senior management. "You have to sell that vision first to major stakeholders. If you don't do that, forget it."
Why? Because AI implementation requires every department to "spend extra time understanding procedures." Without executive sponsorship, department heads will deprioritize it.
Centralization unlocks AI value – scattered knowledge kills it. Tabone's team spent months documenting every policy, procedure, and training material across departments. "We had papers, every department with their procedures, everything scattered around."
Now? "All our policies, procedures, training materials together in one centralized place. When I look back to five months ago, it's actually an amazing feat."
This matters because most retailers try to implement AI on top of fragmented knowledge systems. The AI can't deliver value if information is scattered across SharePoint folders and outdated PDFs.
The payoff is immediate: standardization and multilingual capability. "Foreign workers can listen to everything I would say in English, in Italian, in French, in German, in Hindi." This solves a problem most global retailers face but haven't systematized.
Track progress automatically or you'll never know if training worked. Tabone emphasizes: "We're able to track the progress of every single team member, which is an invaluable tool."
But the real unlock isn't the tracking itself – it's what it enables. Tabone's dream: "An employee calling in sick opens their app, says 'I'm sick,' and the agent immediately contacts and organizes a replacement whilst contacting management and payroll." That level of automation only works if you have real-time data on who's completed which training.
What to do about this:
→ Secure executive commitment before evaluating any AI vendors. Schedule a working session with Board and senior leadership to define: What business problem are we solving? What does success look like? Document this as a charter before evaluating tools.
→ Centralize your knowledge before implementing AI. Conduct a 30-day knowledge audit across HR, training, and operations. Identify all scattered policies and procedures. Assign owners to consolidate this content.
→ Pilot with diverse store profiles, not just your best performers. Select 3-5 pilot locations spanning the performance spectrum. Run the pilot for 4-6 weeks, gather feedback, make adjustments, then commit to full rollout.
→ Build automatic progress tracking from day one. Require completion quizzes for every training module. Create dashboards showing completion rates by store and role. Set triggers for non-completion.
→ Frame transformation as enabling higher-value work, not cutting headcount. Measure success by time reallocation to strategic work, not by reducing HR headcount.
3. How a mid-sized retailer turned siloed data into €100K monthly revenue lift
Chatting GPT podcast with Maryrose Lions, Session: AI in the World of Retail with John Clancy, CEO at Galvia (Nov. 11, 2025)
Proven framework for mid-market retailers to extract value from existing data systems, with specific examples of revenue uplifts and actionable basket analysis strategies.
For: CDOs/CMOs in mid-market retail, digital transformation leaders in specialty/grocery retail, retail analytics directors, heads of e-commerce for regional chains, PE-backed retail operators
Background: John Clancy built Galvia to solve a problem most retail AI companies ignore: the 8,500 UK retailers with 5-20 stores who have plenty of data but no dedicated AI division. These businesses – outdoor living chains, pet food stores, specialty grocers – grew organically and now run 10-20 disconnected systems. They can't predict what's happening in 3 months. They don't know which customers are about to churn. They can't see basket patterns across online and in-store.
TLDR:
Mid-market retailers (5-20 stores, 100-300 employees) sit on disconnected gold mines – Galvia's ML models ingest feeds from 10-20 systems nightly, creating unified predictive views without replacing infrastructure. One retailer discovered €11K/week in Google ads attracted 2,800 customers in January, but only 100 remained active – redirecting spend to churning customers drove €102K monthly uplift.
Basket analysis reveals hidden revenue in traffic you're already paying for – a Manchester outdoor living retailer found 1,800 café visitors in two weeks with only 1% store conversion. Upskilling café staff to cross-sell while customers waited drove "hundreds of thousands to the bottom line."
Pet Stop discovered zero cross-selling during €40-50 grooming appointments despite customers physically present with 10-minute wait times.
Here's the framework many retailers miss:
You don't need to replace your systems – you need to connect them with AI plumbing. Clancy is direct about the mid-market reality: "These companies have grown organically, but might have 10 to 20 systems that underpin their business. It's very hard for them to get a view of what is actually happening across the business."
Most AI vendors want to rip out your tech stack and sell integrated replacements = $500K-$2M projects taking 18 months. Galvia's approach: "Keep using the systems you're using. We don't replace anything. We're like the AI plumbing." Every night, they ingest feeds from Shopify, warehouse management, CRM – creating a machine learning model "trained on your data. It sits within the four walls of their system. No data ever leaves."
The result is "one view of what's happening in your business" plus predictive intelligence for 3-6 months ahead based on 7 years of pattern data. Implementation takes weeks, not months. For mid-market retailers operating on tight budgets, this is the difference between accessing enterprise-grade AI and waiting until competitors beat you to it.
Basket analysis reveals hidden revenue in traffic you're already paying for – the customers are in your building. One Galvia client runs an outdoor store in Manchester with five restaurants and cafés (1,000-person capacity) plus a pet food store with an in-store café. Basket analysis revealed a stunning gap: "1,800 people visited your café in the past two weeks and you had 1% spend money."
The café staff thought they worked in the café. They didn't realize they were part of the overall brand experience. Simple upskilling – teaching them what's on sale, suggesting relevant products during wait times – drove 3-5% basket increases worth "hundreds of thousands to your bottom line."
Pet Stop in Ireland (14 stores, 300 staff) had the same problem with grooming services. Customers spend €40-50 on grooming with 10-minute wait times – captive, high-intent shoppers – but "they had no upsales, no cross sales." Clancy's AI suggested bundling dog food with collars during appointments. The insight: "You have somebody in your store physically that is a high-end spend. Why don't you sell more?"
This isn't sophisticated marketing theory. It's operationalizing obvious opportunities that stay invisible because data lives in silos. When your grooming system doesn't talk to your inventory system doesn't talk to your loyalty program, you can't see these patterns. Galvia makes the invisible visible.
Most marketing spend attracts one-time buyers, not customers – connected data reveals this before you waste millions. A Galvia client spent €11,000 weekly on Google ads. Klaviyo reported good engagement. Shopify showed decent conversions. The campaigns looked successful.
Then the CEO asked: "Where are those clients now that I spent all this money in January 2024?" Galvia's AI connected the dots: "You got 2,800 odd clients in January 24. 100 are with you today. Those campaigns are not working. You're attracting the wrong customer profile."
The fix: "You have 140,000 people here that are going to churn or have churned. Target these with the campaign." Result: 4.4% revenue uplift, €102,000 in one month. Same ad budget. Different targeting based on retention data, not acquisition data.
This is why disconnected systems destroy value. Your ad platform optimizes for conversions. Your CRM tracks engagement. Your e-commerce platform reports sales. But none of them answer: "Are these customers still here 90 days later?" That question requires connecting three systems most retailers run independently.
What to do About This:
→ Conduct a 30-day basket analysis pilot in your highest-traffic location. Don't start with AI, start with questions. Pull data for customers who visited specific departments (café, grooming, service desk) and cross-reference with purchase data. Look for: What % of service visitors buy nothing? What's the average basket when they do buy? Which products correlate with service visits?
→ Map your system architecture and identify the "plumbing gaps." List every system that touches customer data: e-commerce, loyalty, CRM, email, inventory, warehouse management. Then ask: Which systems never talk to each other? Where do we manually export/import data? These gaps are where intelligence gets lost.
→ Start measuring "connected thinking" KPIs, not just department KPIs. Stop reporting sales, inventory, and marketing in silos. Start asking: How many customers visit service departments but buy nothing? What's our off-sales rate (customers asking for items not in stock)? What % of acquired customers are still active 90 days later? These cross-functional metrics force connected thinking.
Disclaimer
This newsletter is for informational purposes only and summarizes public sources and podcast discussions at a high level. It is not legal, financial, tax, security, or implementation advice, and it does not endorse any product, vendor, or approach. Retail environments, laws, and technologies change quickly; details may be incomplete or out of date. Always validate requirements, security, data protection, labor, and accessibility implications for your organization, and consult qualified advisors before making decisions or changes. All trademarks and brands are the property of their respective owners.
