Walmart just handed every retail and CPG executive a case study. Most of them are reading it wrong.
The headline says retail media is hot. Walmart generated $6.4 billion in advertising revenue, up 46% year over year. The instinct in the C-suite is predictable: stand up a media network, hire a sales team, start monetizing sponsored search. That is copying the output. It misses the input entirely.
The real lesson sits one layer below the revenue line. Walmart did not get here by selling ads. It got here by turning first-party data into a product instead of treating it as exhaust.
The number everyone quotes, and the one they should
The quotable stat is the top line. The structural stat is the profit mix. In Q4 FY2026, Walmart disclosed that advertising and membership income together accounted for roughly one-third of operating profit, a striking figure for a company historically defined by low-margin retail.
Read that twice. A retailer famous for pennies per transaction now pulls a third of its profit from high-margin services wrapped around the sale. That is not a media network. That is a new business model living inside the old one.
Now the contrarian part. The ad business is still small relative to the enterprise. For Walmart, $6.4 billion in ad revenue only rounded up to 1% of its equivalent total 2025 sales, which was $713 billion. Amazon, by comparison, runs near 8%. The takeaway most executives draw is “do more ads, there is runway.” Wrong instinct. The runway is not the point. The asset under the runway is the point.
Data as a product, not a byproduct
Here is what actually differentiates Walmart, and it has almost nothing to do with banner inventory. Walmart’s differentiation lies in its ability to deliver omnichannel closed-loop measurement at scale. The Walmart ecosystem offers a unique combination of national brick-and-mortar reach, growing online marketplace, first-party transaction data, and a connected TV platform under one roof.
That is data productization in one sentence. Stores, marketplace, transaction history, and screens, all joined and measurable. The Vizio move made it obvious: Walmart now controls the hardware (Vizio TVs), the operating system (SmartCast), and the shopping data, giving it significant control over ad delivery and measurement.
When data is a product, you can sell off it more than once. Ads were just the first SKU. The same asset powers measurement, personalization, supplier collaboration, and increasingly, AI agents. The ad revenue is the receipt, not the recipe.
Most retailers do the opposite. They treat shopper data as a byproduct of moving inventory. It lives in a dozen systems, never joined, never governed, never queryable in production. It is a leak, not a line item. And every quarter that leak runs, the signal decays and someone else compounds.
The CODN is already running
This is where the Cost of Doing Nothing gets concrete. CODN is not a slide. It is the silent line on your P&L that nobody books because nobody is forced to. While you debate a data strategy, your competitor is converting traffic into a measurable, sellable asset.
The market is consolidating fast around the few players who built the plumbing. Amazon Ads commanded 79.7% of the U.S. retail media market in 2025, while Walmart Connect ranked a distant second at 8.0%, more than five times the share of Target Roundel at 1.5%. In 2026, Amazon and Walmart are forecast to capture 89% of incremental retail media spending for the year. Two companies. Eighty-nine cents on every new dollar. That gap did not open because of better creative. It opened because of better data infrastructure, built earlier.
And the broader signal for your boardroom is bigger than media. The future of retail profitability may depend less on selling more units and more on monetizing the ecosystem around those units. Retailers that can convert their traffic, data, and relationships into scalable media and membership businesses will have a structural advantage in earnings power.
So stop benchmarking your retail media network against Walmart’s. Benchmark your data layer against it. Is your first-party data governed, joined, and production-ready, or is it still a quarterly export someone cleans by hand?
Where this goes next
The ad business was phase one of monetizing the asset. Phase two is agentic. Walmart is already there: it is testing ads within its AI chatbot, Sparky, which executives say is already influencing higher basket sizes.
Think about what that requires. An agent that recommends, persuades, and converts inside a conversation only works if the data underneath is a clean, real-time product. The ad model was the warm-up. The agentic model is the main event, and it punishes anyone whose data is still exhaust.
So the question for your 2027 plan is not “should we sell ads.” It is sharper than that: is your first-party data a product yet, or are you still funding someone else’s flywheel with your own shoppers? Walmart already answered. The clock on the rest of us is running.