The Maze: Retailers still judge stores like standalone shops. That made sense when the store did one job: sell what sat inside its four walls. It is a poor way to allocate capital when the same location also picks online orders, extends delivery reach, creates retail-media inventory, pools stock and feeds data back into the customer relationship.
More than 75% of Walmart sales sit in the familiar P&L, but the interesting economics sit outside it. BCG’s illustrative breakdown separates direct sales-floor volume from five network roles. A location that looks thin on its own margin can still support an online order, a faster delivery promise or a denser local audience. The old scorecard can be accurate and still be incomplete. That matters because finance teams do not just rank stores; they decide which locations get labor, inventory, technology and renewal capital.
Fulfillment changes the closure conversation. About 50% of Walmart’s online items are fulfilled from stores, while the same footprint supports under-three-hour delivery to 95% of US households. Those are not store-sales metrics. They are cost-to-serve and coverage advantages. Closing a weak-looking location may remove more value from nearby customers and nodes than its four-wall result reveals. The test is not whether every store must stay open. It is whether the network loss has been measured before the rent saving becomes the answer.
Retail media turns physical density into an audience asset. Walmart’s global advertising business generated $6.4 billion in fiscal 2026. That number is not revenue from any single store, and it should not be assigned to one. It does show why traffic, shopper data and local density now influence investment decisions alongside sales and labor productivity. The store is part shop, part distribution point, part media surface. A local manager cannot run that media business alone, but the local node helps create the traffic and behavioral signal that make it monetizable.
The remaining roles make resilience measurable rather than aspirational. A 5,200-site Walmart and Sam’s Club network can pool stock when one node runs short. Real-time store data can also support tools associated with 35% higher order value in BCG’s example. These are different measures with different denominators; they are not a neat sum. Their shared implication is that management needs role-specific KPIs, not one universal hurdle rate. A fulfillment node should be judged on pick cost, capacity and service levels; a traffic-rich node needs a different mix of conversion, audience and media measures.
Why it matters: The next store-network advantage will not come from declaring every location “omnichannel.” It will come from naming each node’s primary role, measuring the contribution that role makes to the network, and testing closures, openings and technology rollouts against that contribution. A four-wall P&L should remain on the dashboard. It should stop being the whole dashboard.
Sources: BCG, *Retailers Need to Manage Stores as a Portfolio Now. AI Lets You Do That* (August 11, 2026; including its Walmart Exhibit 3 and cited Walmart fiscal-2026 figures).
