Building the AI-ready enterprise: How JD Sports is scaling smarter

AI-ready does not start with AI

For the last year, “agentic commerce” has sounded like a marketing Rorschach test. Some people hear automation and see margin expansion. Others hear platform dependency and see their customer relationship slipping away. The debate has been loud, but the proof has been thin, mostly because the hard part is not the chatbot. The hard part is the enterprise behind it.

That tension sat at the heart of JD Sports’ session on building an AI-ready enterprise. While the panel framed the future as LLM-driven discovery and checkout, Olivier Nachba, Deputy CTO at JD Sports, kept pulling the conversation back to something more sobering and more useful: retail is still a promise business. AI does not replace that promise. It amplifies the consequences when you break it.

JD’s context makes the stakes clear. A global business with multiple brands and a customer base skewing young, JD is dealing with shoppers who expect speed and certainty, and who switch channels without thinking twice. The headline challenge is not “How do we adopt AI?” It is “How do we serve customers anywhere, in any channel, without losing control of margin, stock, and experience?”

Why this research felt different

Plenty of sessions talk about AI like a layer you can add. JD’s perspective treats AI as a pressure test. If you have unfinished omnichannel work, AI will expose it. If your data is fragmented, AI will multiply the confusion. If your operations are brittle, new channels will create new failure modes.

Nachba’s framing was blunt: many retailers still treat omnichannel as a work in progress, splitting digital and store priorities and debating rules as if the customer is not already moving freely between them. In his view, the AI era arrives before that foundational work is complete. Which means the winning move is not to chase whatever is trending, but to harden the core that keeps the retail promise intact.

The credibility comes from the backbone

The clearest anchor in JD’s approach is order management. Not as a software category, but as an operating system for the customer promise.

Nachba shared metrics that make the point tangible. In peak periods, JD can see 100,000 orders in an hour. The system response times he cited sit around 100 milliseconds. And the metric he cares about is cancellation rate, because nothing breaks trust faster than selling what you cannot fulfill.

This is the part many marketers gloss over. “AI-ready” is often described as content, personalization, and new experiences. JD’s version starts with orchestration: knowing where stock is, deciding how to route orders, and setting delivery expectations that hold up under stress. When inventory is scattered across DCs and stores, and the “lonely pair of shoes” might be sitting in one location, the business advantage goes to whoever can calculate availability and promise delivery fastest, while still protecting margin.

GEO and agentic commerce: two different problems

Nachba separated two ideas that often get lumped together.

GEO, as he described it, is about being found in an LLM-shaped internet. If a customer asks for a specific product in a specific city, JD wants the model to surface the right answer at both global and local levels. That is not just SEO with a new label. It is a new communication pattern, where retailers need enough accurate, targeted content at scale to remain visible.

Agentic commerce is different. That is the purchase itself moving into interfaces like ChatGPT, Gemini, or Microsoft’s ecosystem. For JD, it is simply another channel. The customer may want to browse and buy without visiting a traditional site, and JD’s job is to be present where the customer chooses to transact.

The strategic takeaway is that “visibility” and “transaction” do not share the same readiness path. GEO is a content and data discipline. Agentic commerce is a fulfillment and promise discipline. Both fail if your data is not clean.

The hidden constraint: organizational alignment

The toughest ROI conversation, Nachba suggested, is not about software value. It is about internal incentives.

Order management sits across supply chain, digital, and stores, and each group can optimize for its own P&L. Digital might push volume through DCs. Stores might prefer ship-from-store. Without alignment, the system becomes a political battleground, not a customer engine.

His solution was simple but demanding: restart from the customer. Build the operating rules around what the customer expects, then align stakeholders to deliver it. Amazon was his shorthand example, not because of web design, but because the promise is reliable. When it says 24 hours, it means 24 hours.

Creative and experience implications: the store still matters most

Despite the AI focus, Nachba returned repeatedly to the store. Roughly 80% of sales are still in-store, he noted, and the brand experience is what brings customers back. Rewards programs and digital perks do not compensate for a weak experience.

The next challenge is that customers will walk in with more information. If they have already searched via an LLM, store colleagues need the tools and training to match that confidence, including fulfilling an order even when the item is not physically in that location. That is where order management becomes a frontline experience tool, not just a logistics system.

What marketers should take from JD’s AI readiness

JD’s message is not anti-AI. It is anti-theater.

They are using AI to give teams “superpowers” to produce more targeted content for GEO. They are thinking about being a “five-star service” in new LLM-driven channels to avoid being deprioritized. They are embracing test-and-learn, including A/B testing fulfillment and sourcing rules, to replace endless meetings with measurable outcomes.

But the core discipline remains operational truth. Clean data. A source of truth for products and stock. Systems that can make and keep a promise.

For senior marketers, the implication is clear: the future funnel starts earlier and becomes more conditional. Availability and delivery promises are moving up the funnel, even into ads, as one example discussed on stage showed, where inventory availability was connected to an ad platform to avoid promoting products that cannot be fulfilled.

AI will change discovery, decisioning, and checkout. But the winners will not be the brands with the slickest demo. They will be the businesses that treat AI as a forcing function to fix the fundamentals, so every new channel becomes an opportunity instead of a risk.