■ CRITICAL RISK ■ Retail
Yes, and in most big-box chains it effectively already has. Between RFID tags, shelf-scanning robots, and perpetual inventory software, the dedicated human counter is now the backup plan, not the plan.
“RFID counts faster than you can blink.”
Our AI replacement risk score — how we score jobs
Inventory counting is exactly what it sounds like: walking aisles or warehouse racks with a scanner (or, in the old days, a clipboard and a ticket), tallying what's physically on the shelf, and reconciling it against what the system says should be there. The job clusters around overnight cycle counts, quarterly full counts, and the miserable annual physical inventory where a crew descends on a store at 2 AM. Accuracy is the whole game — a miscount cascades into bad reorders, phantom stockouts, and angry store managers.
The trouble is that every piece of that workflow has been automated from a different direction at once. RFID tags let a handheld wand — or a fixed ceiling reader — count an entire apparel department in minutes without touching a single item. Shelf-scanning robots roll the aisles nightly photographing every facing. Computer vision on ordinary security cameras is starting to do the same job with no robot at all. And perpetual inventory systems tied to point-of-sale data mean the count is continuously estimated rather than periodically discovered. Third-party inventory services still hire crews, but the crews get smaller and the tech does more of the actual counting each year.
What survives is the exception handling: the mislabeled pallet, the shrinkage investigation, the backroom chaos no camera can parse, and small retailers who can't justify RFID tagging every SKU. Someone still has to physically find the box the system swears exists. But 'resolving count discrepancies' is a task inside someone else's job — a stock associate, a loss-prevention analyst — not a career of its own. Our risk score of 91 reflects a role where the core deliverable, the number, is now produced by hardware.
Automatability: our editorial assessment of current and near-term AI capability
This one isn't a forecast — it's a postmortem. RFID rollouts at major apparel and general-merchandise retailers have been shrinking count crews for over a decade, and shelf-scanning robots and camera-based inventory systems are eating what's left right now. Expect dedicated counting work to keep contracting through the late 2020s until it survives mainly in small independent retail and as an exception-handling task folded into other roles.
As a stepping stone, yes — third-party count services and retailers still hire, especially for overnight and seasonal work. As a career, no. RFID and shelf-scanning automation are shrinking crew sizes year over year, and the remaining work is increasingly absorbed into general stock-associate roles rather than existing as a standalone occupation.
Three things at once: RFID tags that let entire departments be counted in minutes, autonomous shelf-scanning robots that photograph aisles nightly, and perpetual inventory software that estimates stock continuously from sales and receiving data. Computer vision running on ordinary store cameras is the newest entrant, and it requires no special hardware on the products at all.
Climb the stack. The counting is automated, but the systems that do it need operators, and the numbers they produce need interpreters. Inventory control specialists, WMS administrators, and loss-prevention analysts all build directly on counting experience. Even a short certification in supply chain basics or a specific WMS platform moves you from the automated side of the job to the automating side.
For a while, yes. RFID tagging every item only pencils out at scale, so independent shops and businesses with messy, low-volume stock will keep counting by hand or with basic barcode scanners. But that's a shrinking island, and it's usually the owner or existing staff doing the counting — not a hired specialist.