The writing was on the wall — or rather, missing from the shelf — from Day One.
Last September, Starbucks Corp. proudly dropped a promotional video showcasing its shiny new Automated Counting system. Designed with computer vision startup NomadGo, the tool was supposed to liberate baristas from tedious clipboard tallies. Instead, the video captured an awkward glitch: the artificial intelligence (AI) completely failed to recognize a bottle of peppermint syrup while scanning adjacent stock.
It was a quiet omen of a noisy retail reality. Nine months later, Starbucks has officially retired the program across North American stores, according to internal documents reviewed by Reuters.
For front-line workers, the decision was cause for celebration. “Thank you for trusting the partners over unreliable spatial recognition to handle these counts,” one employee wrote on an internal feedback channel. Another added, “The thought behind it was great, but the execution was proving difficult.”
The app’s premise was simple: baristas would hold a tablet up to supply shelves, and the software would use cameras and LiDAR data to track inventory. The goal was to eliminate persistent product shortages, which was a major headache for CEO Brian Niccol, who took the helm in late 2024 to spearhead the “Back to Starbucks” turnaround campaign.
But retail floors are chaotic environments, exposing AI’s brittleness in ways pristine tech labs never could. The computer vision struggled with basic tasks, frequently miscounting items or confusing similar-looking dairy products.
In a coffee shop, a technical hiccup creates a supply chain nightmare. When an algorithm cannot distinguish between whole milk and oat milk, it triggers automatic reorders for products already flooding the backroom, while leaving the store completely dry on what customers actually want. For baristas, an automation tool that is wrong even 10% of the time doesn’t save labor; it doubles it, forcing humans to audit the machine.
Starbucks framed the cancellation as a tactical move. In a statement, the company said it has “moved to a single, consistent process across all inventory counts” to focus on consistency and execution at scale, while being disciplined about where automation truly adds value.
Niccol remains under intense investor pressure to revive profitability. While the coffee giant posted its strongest quarterly sales growth in two and a half years in April, pushing its stock up 24% in 2026, North American operating margins have plummeted to 9.9% from 18% two years ago.
To fix the fractured supply chain, Niccol is shifting focus away from the shelves and toward logistics, hiring fresh supply chain executives and promising daily store replenishments by the end of 2026. “Our goal is simple — if it’s on the menu, customers should be able to order it,” the company said in a statement.
Starbucks is far from the only fast-food giant to realize that the automated future is harder to engineer than it looks. McDonald’s Corp. recently scrapped its drive-thru voice AI partnership with IBM Corp., Taco Bell slowed its own voice-tech rollout, and a major Pizza Hut franchisee blamed an AI ordering system for $100 million in lost sales.
Niccol isn’t abandoning technology entirely; Starbucks is still deploying AI to sequence digital orders and assist baristas behind the counter. But as the automated inventory experiment proves, the best technology in retail is the kind that works invisibly in the background — not the kind that leaves baristas crying over miscounted milk.
