The Supply Chain Runs at 98%. The Shelf Runs at 90%.

Walk any store with a scorecard in hand and the same conversation happens. The system says the item is in stock. The buyer’s report says the item is in stock. The shelf says otherwise. Everyone in the conversation is telling the truth, because they are measuring three different things.

Availability comes in layers, and the layers are routinely treated as interchangeable when they are not. Supply chain availability means stock exists somewhere in the network, in a distribution center, in transit, on order. In-store availability means stock is physically inside the building, which includes the back room, the top stock, and the pallet nobody has worked yet. On-shelf availability means the product is on the shelf, in position, shoppable at the moment a customer reaches for it. Only the third layer produces revenue, and the third layer is the one most measurement systems see least clearly.

I have spent more than four decades on this problem from every seat that touches it. I was part of the team at Procter & Gamble that helped Walmart develop what became Retail Link, and I later led Walmart’s RFID and on-shelf availability programs. I led the Supplier Portal Allowing Retail Coverage (SPARC) team, which gave suppliers the ability to measure and improve on-shelf availability at Walmart. The lesson that held across all of it is that the industry’s hardest availability failures live in the last fifty feet, between the back room and the shelf, and that most measurement systems stop just before the failure starts.

The research community documented the shape of this years ago. ECR Europe’s Optimal Shelf Availability research, conducted with Roland Berger, found service levels of 98 to 99% from the manufacturer’s warehouse through to the retailer’s stockroom, collapsing to 90 to 93% over the final meters from the stockroom to the shelf. The industry hit its availability targets almost everywhere except the last place the product travels, which is the only place the shopper ever visits. And the problem has persisted at scale: NielsenIQ reported that 7.4% of U.S. CPG sales went unrealized due to out-of-stock and out-of-shelf items in 2021, an $82 billion cost in that single year.

The Bill for the Last Fifty Feet

The cost of that final gap is not small. IHL Group’s September 2025 research puts the global cost of inventory distortion, the combined toll of out-of-stocks and overstocks, at $1.73 trillion annually, equal to 6.5% of global retail sales, and the firm’s tracking attributes $1.2 trillion of that total to out-of-stocks. Research published by ECR Retail Loss in 2025 found that 43% of shelf out-of-stock incidents result in less money in the till once substitutions are netted out, with a Walmart Data Ventures study evidencing the value lost even when shoppers do substitute. And the ECR shopper research describes how patience runs out: by the third time a shopper hits a gap on the same item, the probability they switch stores reaches 70%.

The damage concentrates in the worst possible windows. Research by Gruen and Corsten across the fast-moving consumer goods industry found average out-of-stock rates around 8%, a figure ECR Retail Loss’s 2025 analysis still found sitting at 8.6%, with promoted items running at roughly double the average. Read that against how trade investment works. The brand funds the promotion, negotiates the feature, activates the media, and drives the traffic, and availability is most likely to fail during exactly those days, when conversion was the entire point of the spend. A brand that measures only supply chain availability will conclude the promotion underperformed. The promotion may have performed fine. The shelf did not.

The Shelf Now Fails Two Shoppers at Once

For most of retail history, an empty shelf disappointed one person at a time. Online fulfillment from stores changed that arithmetic. The same gap now fails the customer who walks the aisle and the picker filling an online order on behalf of a customer who is not there, and the picker’s version of the failure is worse, because it ends in a substitution decision made by a stranger or an item refunded off the order entirely.

Harvard Business Review documented what fixing the picker’s version is worth. It reported an Instacart experiment in which customers steered toward delivery windows with better stock positions increased average daily spending by 4.6%, with fewer items replaced or refunded. That is a measurable revenue difference produced by nothing except availability at the moment of picking. Every retailer running store fulfillment now generates the same signal at scale, whether or not anyone is reading it: pick rates, substitution rates, and refund rates are a continuous, order-by-order audit of on-shelf availability that did not exist when the industry’s measurement habits were formed. The industry is starting to act on that idea, and Conversations on Retail covered its newest expression earlier this week, as Instacart moved to turn the person already in the aisle into its availability signal.

Measure the Third Layer, Then Manage It Together

The fix begins with honesty about which layer a metric describes. A perpetual inventory figure is not an on-shelf number. A distribution center fill rate is not an on-shelf number. On-shelf availability is measured at the shelf, from the shopper’s side of it, whether the measurement arrives through store audits, computer vision, RFID, or the fulfillment signals already flowing through the picking operation.

The opportunity then belongs to everyone who touches the shelf, because no single party controls the last fifty feet. Retailers own the replenishment processes and labor models that move product those final steps. Suppliers own the forecast quality, the case pack decisions, and the promotional volumes that determine what those processes have to absorb. Brokers and field teams are often the only people physically standing at the shelf with authority to fix what they see. When those parties manage to a shared on-shelf number in joint planning, the gap closes; when each manages its own layer, the gap persists, and each party’s numbers say the problem belongs to someone else.

The ECR researchers called the moment a shopper reaches for a product the moment of truth, and their finding was that everything upstream of that moment already performed at 98. The shopper never sees the 98. They see a gap or they see the product, and they decide from there.

View Profile

Mike Graen

With 42 years of experience working for Procter & Gamble, Walmart, and CROSSMARK, Mike Graen has seen the retail industry from every conceivable angle.

From helping develop what would become Retail Link to investigating and leveraging the potential of shelf-scanning robots and RFID tech in stores, Mike has made a career out of using technology to solve business problems.

Mike partners with retailers, consumer product companies, and solution providers from all over the world, leveraging his reach, relationships, and influence to help teams deliver product availability while driving strategic alignment between retailers, their supplier partners, and the solution providers who enable them both.