Retail's Quiet Winners: Why success when scaling AI starts small
While McDonald’s tested AI drive-thru ordering in over 100 locations before abandoning the entire partnership, Decathlon piloted their loyalty programme in one market, proved profitability, then gradually expanded to 69 countries. The difference was discipline.
In our first article, we identified three characteristics of retail’s AI winners. Our second article examined the second of these, the people-first investment and enablement approach. This final piece explores the third characteristic: proving value before scaling AI through disciplined pilots.
The Pressure to Scale
Every retailer faces the same contradiction: boards demand transformation while pilots feel too small, yet most AI initiatives never reach maturity. Whilst the pressure is all around. Competitors announcing AI strategies. Vendors promising enterprise-wide evolution. The news cycle constantly amplifying the relentless march of AI technologies. The result? Pilot proliferation that dilutes talent, investment and focus, fragments infrastructure and prevents initiatives from accumulating the learning needed to succeed.
The Framework
Yet the quiet winners resist this pressure through disciplined progression. As explored in our first article, Decathlon’s loyalty programme transformation illustrates this approach. Starting with a pilot in Poland with one clear goal: prove the programme would generate profit and it generated engagement.
The pilot revealed actionable patterns that informed inventory decisions, staff scheduling and personalised communications. From Poland, Decathlon expanded to three test markets, refined the model, then gradually rolled out to their entire 69-country network.
Measurement discipline is vital. Lowe’s tracked $1M+ in operational savings within eight months, including $325,000 in administrative labour and 434,000+ hours of automated schedule changes.
This measurement rigour creates two advantages. First, it generates evidence for continued investment, as generally CFOs will fund proven ROI, but not “future indirect value” promises. Second, it reveals what’s not working fast enough to pivot before wasting resources.
What Pilots Actually Prove
Successful pilots don’t guarantee production success. They prove technical feasibility in controlled environments, not scalability, organisational readiness or real-world robustness.
Effective pilots test three things beyond technology:
Workflow integration - How does this AI improve existing processes?
User adoption - Do people use it without coercion? Is the value obvious?
Value clarity - Can everyone articulate why this matters?
The Compound Value of Depth Over Breadth
Scattered pilots never reach critical mass. Deep implementation creates compound advantages. Myer built Oracle Retail planning, deployed 3,600 mobility devices, created four operational applications and launched M-Metrics. Each component strengthened the others through shared technical foundations and organisational learning.
Nike’s distribution centres integrate supply chain AI, RFID tracking, demand-sensing platforms and over 1,000 collaborative robots, combining to deliver end-to-end operational excellence supporting 44% direct-to-consumer channel growth.
Most interestingly, purchased AI tools succeed 67% of the time while internal builds succeed only 33% as often. Expert partners bring concentrated learning. Internal builds that are spread thin across the business, accumulate learning slowly, if at all.
The Three Questions Before Scaling
Can we articulate why it worked? Which insights drove which outcomes? Causal understanding, not just correlation, enables confident expansion.
Have we built the organisational capability to support it? Scaling requires training programmes, support structures, and feedback loops. Technology works because the organisation can sustain it.
Does the ROI improve or degrade at scale? Some initiatives spread fixed costs. Others hit complexity that erodes returns.
If you can’t answer all three confidently, you’re not ready to scale. Staying small longer becomes a competitive advantage.
Why This Is So Hard
The discipline of proving before scaling runs counter to organisational instinct. Boards demand visible progress. Investors want scale stories. Internal politics pressure teams to show enterprise-wide impact. The temptation to declare victory after a successful pilot and immediately scale is overwhelming.
The quiet winners resist through three commitments: measurement discipline with outcomes determined before launch, organisational honesty that addresses problems rather than rationalising them and patient investment that keeps projects operational for years, not quarters.
The Strategic Choice
As with all technology initiatives, this choice remains: do you scale before proof or prove value before scaling?
The difference is a disciplined end to end:
Identify a specific use case - A focused problem with measurable impact.
Quantify the value before starting - What ROI justifies continued investment?
Design a focused pilot - Set a clear goal. Resist scope creep.
Measure relentlessly - Did it deliver the quantified value? Track with CFO rigour.
Make a clear decision - Pass or fail. Scale or stop. Clear gates, honest assessment, action on evidence.
This transforms businesses through improvements that compound over time. The work isn’t glamorous but it ensures the quiet winners continue their methodical path to success.