An RFID pilot can produce a satisfying set of results. The count moves quickly, inventory accuracy improves and the project team leaves with enough evidence to recommend a wider deployment. Months later, the same technology may be producing uneven results across the store network, with no single failure large enough to explain what changed.
This pattern does not mean the pilot was misleading. A pilot is a protected environment by design. The included merchandise receives closer inspection, the participating employees know the project matters and technical support remains within easy reach. A rollout has to perform under a different set of conditions. Inventory arrives on a busy afternoon. A return is placed back on the floor without its original tag. An employee conducts a cycle count while also answering questions from customers. The technology has moved from a project into the working life of the business, where small deviations have time to accumulate.
Writing after EuroShop 2026, Impinj Senior Vice President Suresh Palliparambil described the industry’s current position plainly: “The industry has moved beyond pilots. The focus now is on optimization.” That observation captures the work facing many retailers today. RFID has already demonstrated its capabilities. The harder task is building a repeatable operation around those capabilities.
A Pilot Receives Treatment a Rollout Cannot Depend On
Table of Contents
During a pilot, questionable tags can be replaced before the reader ever enters the department. Product records may be reviewed by people who know where past errors tend to hide. The counting route is discussed in detail, and someone from the project team is usually available when the first unexplained variance appears. Those precautions are appropriate because the company is trying to learn how the system behaves.
That degree of attention becomes difficult to maintain as more locations enter the program. A store manager may receive a training document and a shipment of equipment, then be expected to conduct the next count with the employees already scheduled that day. Older inventory may still be present from before the tagging standard was established. The process now has to work without the concentrated supervision that helped the pilot succeed.
This is why the first phase of a rollout should still be treated as a period of observation. Expanding from one test store to ten stores may reveal more than moving immediately to one hundred. The additional locations introduce differences in layout, staffing and merchandise flow while the implementation team is still close enough to understand the source of a problem. A slower beginning can prevent the company from distributing the same unclear procedure across an entire network.
Find Out How Much Inventory RFID Can See
When count performance begins to slip, the reader is an easy place to look. Employees may change the power setting or scan the department again from a different angle. Those steps have value when the tags are present and functioning. They accomplish very little when a portion of the merchandise never entered the RFID environment. Tag saturation measures how much of the in-scope inventory carries a usable RFID tag. A location can report a strong read rate while still producing an incomplete count because read rate begins with the tags available to the device. Merchandise without a tag falls outside that measurement. The reader did not miss it in the technical sense. The item was never available to be read.
A storewide saturation percentage may also provide false reassurance. Consider a retailer with excellent coverage in its newest and fastest-moving departments but weak coverage among older inventory elsewhere in the building. The location’s average can appear healthy while the same category produces discrepancies week after week. Breaking saturation down by department or merchandise source makes the result more useful because it shows where the tagging process has started to fray. Returns deserve separate attention during this review. Products may come back without their original hangtags, then reenter available inventory before anyone checks for a functioning RFID label. The missing tag does not create an immediate problem at the service desk. It appears later, during a count conducted by someone who has no reason to connect the variance with a return from several days earlier.
Datascan identifies saturation audits and tag-up services among its RFID specialties. Those services address the portion of an implementation that can be overlooked when a retailer concentrates primarily on readers and software. A saturation audit gives the business a clearer starting point before it begins changing equipment settings or questioning the accuracy of every count.
Product Data Travels With the Rollout
A tag can be read perfectly and still create incorrect inventory information. The encoded identifier may point to an outdated product record, or two similar items may have been confused when the association was created. From the reader’s perspective, the transaction was successful. The device received a signal and delivered it to the software without knowing that the business expected a different product.
These errors become more consequential at scale because product data is commonly shared across locations. A bad association that affected a few units during the pilot can begin appearing anywhere the product is carried. Store employees may correct the inventory quantity locally, only to see the variance return during the next cycle count because the shared record remains unchanged. The adjustment fixes the visible number while preserving the source.
The product master should therefore be treated as part of the RFID implementation rather than background information supplied to it. Retail employees can contribute important context here. A receiving manager may know that a supplier changed its packaging without clearly separating the old product identifier from the new one. An associate may know that two products are regularly confused because their labels look nearly identical. These details rarely appear in a technical report, yet they can explain why a clean RFID read continues producing the wrong operational result.
Returns Show How Mature the Program Has Become
New merchandise usually enters through a controlled receiving process. Returns arrive in less predictable condition. Packaging has been opened, labels may be missing and the employee processing the transaction may be working from limited information. The item then moves back into sellable inventory through a path that the original RFID plan may have given only a sentence or two.
One common problem develops when a replacement label is applied without locating the original RFID tag. The physical inventory contains one item, while the next count receives two valid identifiers. The overage may be adjusted without anyone finding the duplicate, allowing the same false unit to return during every later count. The reverse can happen when an item reaches the floor without being retagged. It remains available for purchase while disappearing from RFID counts, creating a shortage that appears to be caused by scanning performance. The count team may search the department several times before anyone traces the item back to the returns process.
A workable procedure does not need to turn every return into a technical inspection. Employees need a dependable way to confirm tag status, apply a replacement when required and remove loose labels from the counting environment. An active tag left in a drawer can keep an item appearing in stock after the merchandise has left the building, sending employees on a search for a product that exists only as a radio response.
Counting Routes Need Room for Judgment
RFID removes the need to scan each barcode individually, but the operator’s path still shapes the count. Merchandise stored behind a dense fixture may require a slower pass. Items awaiting customer pickup may sit outside the department where the system expects to find them. A route that works in one store can leave gaps in another because the physical environments do not match closely enough.
Standardization is still useful. Employees should know where the count begins and what spaces belong to each zone. The problem comes when a route is taught as a walking pattern instead of a method for achieving coverage. An employee can follow the prescribed aisle perfectly while passing several places where merchandise has accumulated outside its usual home.
Training should explain the reasoning behind the route. When employees understand how distance, orientation and product density affect coverage, they can adjust without abandoning consistency. They become more capable of recognizing a weak pass before an exception report forces them to return to the same area.
Observation usually reveals more than another revision to the training manual. Allow an employee to complete a count without being corrected at every turn, then review the path afterward. The spaces they skipped will show where the procedure and the store’s daily reality have parted company.
Integration Decides Where the Data Goes Next
A fast count has limited business value when the information remains inside the RFID platform. The quantity has to reach the systems responsible for replenishment, fulfillment or financial reporting at a time when those systems can still use it. Delays and conflicting ownership rules can leave employees choosing between two inventory figures without understanding why they differ.
The integration plan needs a clear answer to a deceptively simple question: which system owns the inventory quantity after the count? If RFID results are treated as recommendations, someone must review and approve the changes. If adjustments are posted automatically, the company needs rules for handling transactions that occurred while the count was underway. Neither method is inherently correct for every retailer, but ambiguity creates an environment where unexplained differences remain open for too long.
In 2024, Datascan joined the Auburn University RFID Lab Advisory Board, bringing the perspective of an inventory implementation provider into a research organization focused on RFID’s business case and technical execution. In the announcement, Datascan Global Director of RFID Solutions Luke Kosman spoke of “helping retailers pencil a return on their technology and operations investments.” The wording is useful because it places implementation decisions inside a business calculation rather than treating integration as a technical afterthought.
Companies moving from a pilot into a wider deployment can review Datascan’s RFID inventory solutions, which include item-level software designed to capture and distribute RFID data within a retailer’s technology environment. The ability to collect large quantities of information matters, but the larger operational gain comes when that information reaches the right system soon enough to influence the next decision. (Datascan)
Exception Handling Is Where the Rollout Earns Its Keep
A pilot report usually emphasizes the percentage of inventory captured. A mature program places greater weight on what happens to the smaller portion that remains unresolved. Those exceptions are where product-data errors, weak tagging procedures and timing differences begin to reveal themselves.
A full rescan may feel like the safest response, though it can reproduce the same blind spots. A more useful second pass begins with the importance of the discrepancy. One missing item may require immediate attention because it has already been promised to a customer. Several units of slow-moving merchandise may allow time for a review of recent transactions before anyone walks the floor again.
The investigation should leave behind more than a corrected quantity. Recording the likely cause allows patterns to develop across counts. A recurring problem tied to returned merchandise calls for a different response from a problem limited to one supplier. Over time, the exception history can tell the company where inventory accuracy begins to deteriorate, which gives management something more useful than a long record of adjustments.
This is also where staffing decisions become important. Some retailers can manage routine cycle counts internally once the process is established, while larger conversions or validation counts may place too much pressure on the existing team. The decision should consider the cost of pulling experienced employees away from their regular responsibilities as carefully as it considers the cost of outside support.
The Business Case Should Survive Ordinary Conditions
A pilot’s return calculation can lean heavily on count speed because the difference is visible immediately. The old process required a certain number of people for a certain number of hours, and RFID completed the work sooner. That savings matters, though it may represent only part of the value. More dependable inventory can reduce the time employees spend searching for products that are no longer present. It can support better replenishment decisions and give an ecommerce team greater confidence when exposing store inventory online. These gains are harder to measure unless the company establishes a baseline before the pilot begins, which is why the business case should be designed before the most promising results arrive.
Decathlon’s RFID program offers a useful view of the connection between inventory visibility and the customer. Hervé D’Halluin, the company’s RFID and interactive sport products leader, explained that “RFID helps satisfy customers by allowing them to see if the bike they want is available.” The company’s experience also connected stronger availability data with avoiding unnecessary overstock, showing how the value can reach both sides of the inventory equation. (gs1us.org)
The costs deserve the same breadth of attention. Readers and tags are easy to place in a budget because they arrive with prices attached. Training, exception review and maintenance of the surrounding data require time that may be spread across several departments. A return calculation becomes more credible when those less visible costs are included from the beginning rather than discovered after expansion.
Bill Hardgrave, founder of the Auburn RFID Lab, described the Lab’s approach during its 2025 conference: “It wasn’t about technology for technology’s sake. Rather, it was about the use cases.” That standard applies directly to RFID expansion. The technology earns its place through the operating problems it resolves and the decisions it improves. (Harbert College of Business)
Scale Only What the Business Can Repeat
A successful rollout does not require every location to behave identically. It requires the company to understand which parts of the process need firm standards and where local conditions require judgment. Tag association rules should remain consistent. Store layouts will still influence the way employees move through a count.
Expansion can pause briefly at each stage without losing momentum. A group of new stores should complete enough count cycles to show where problems repeat before the next group is added. The purpose of that pause is not to wait for perfection. It gives the implementation team enough evidence to separate one unusual store problem from a procedure that will create trouble everywhere. The rollout is becoming durable when store employees can explain the count process in their own words, exception causes are becoming easier to identify and inventory accuracy holds without constant attention from the project team. At that point, the company has built more than a successful RFID installation. It has created an operating practice that can continue improving after the excitement of the pilot has faded.
