How does enterprise RFID asset management land? Phased implementation path
Last year we helped a precision manufacturer count its assets. The finance team spent two full weeks checking items one by one with barcode scanners in teams of four, and the result still differed from the books by over 300 units. Someone joked it would soon cost more to count the missing assets than to replace them. After switching to RFID, two people counted the same number of assets in half a day. Data synced in real time and the error rate was nearly zero. It showed me that RFID asset management is not just a matter of attaching tags. Deployment needs to happen in stages, which I'll describe below.
Phase one: mapping the family, and not hurried to label.
And many companies come up and buy labels, they buy readers, and it turns out that it's useless to put them on, and the count is not right. The first thing that happens is you don't get it right. This stage answers three questions: where your assets are moving, who's in charge, what's the easiest to lose.
Take an electronics contract manufacturer with CNY 500 million in annual revenue. It had more than 20,000 fixtures, molds, and test instruments on its production lines. It used to record them in Excel, and every stocktake required a full day of downtime. Before introducing RFID, we recommended taking two weeks to classify the assets: Class A (worth over CNY 50,000 and critical to production), Class B (mid-value and frequently used), and Class C (low-value consumables). Applying RFID tags only to Class A and B assets cut tag costs by 40%.
One of the things to do at this stage is field surveys. The RFID signal is metal, it's liquid-drive, and metal shelves, and water tanks in the workshop are going to drop reading rates below 60%. The survey is going to take place at key locations, with a live-reading tag rate, and a record of the blind area, which will provide a basis for subsequent antenna layout. The survey tools typically use portable readers plus several test labels, and can be concluded half a day.
The RFID-minority UHF and HF (UHF read distance, suitable for a mass inventory; HF read distance, appropriate for single precision). More than 90% of the asset management scene is between 860-960 MHz. The labels also contain various types of seals, which are placed on the metal surface to select an anti-metal label, and normal card tags are largely unreadable on metals.
Phase II: small-scale pilot, running the closed loop
Before the whole plant is opened, a region must be selected for piloting. It is suggested that it should start with a warehouse or an independent workshop, and be between 500 and 2000 items of assets. The pilot is not intended to prove that RFID works well, but to expose problems.
The problems that are common during the pilot period are of these types:
- The label is not in the right position, it's blocked by the device. It has a 70% reading rate.
- The process of the outing is still unchanged, assets are labelled and the system is not known, and it becomes "detached."
- reader antennas are re-loading, not in the remote area.
- Backstage software and ERP fields not match. Data sync error
We were piloting a car parts manufacturer, and the initial reading rate was 82% off. The site ran a high-frequency welding device in the workshop, which generated electromagnetic noise that put the RFID signal under pressure. Later, the reader antenna was moved to 3 metres from the welding position, and a suction material was added to the bottom of the label, and the read rate stabilized to 99.2%. This problem could not be detected on site, and paper alone was not planned.
The pilot typically lasts 4–8 weeks and tests the full workflow: tagging, goods receipt, checkout, return, and inventory counting. At this stage, verify three things in the back-end software: bulk inventory speed, exception alerts, and data interfaces with existing ERP and MES systems. A benchmark is 50–200 tag reads per second per fixed reader and a complete count of 10,000 assets within one hour.
Phase III: Quantities and side-slippers
The most likely mistake of this move is "one size fits all" , with all the assets being laid out at once. On a scale, coordination on the ground doubles, and problems are not over.
A practical approach is to roll out by area: start with the workshop in Building A, move on to Building B, and finish with the warehouse. Allow two to three weeks between batches to monitor results and resolve outstanding issues. At a plant producing 100,000 units a year, three to four batches over three to six months is a reasonable schedule.
The one thing that you need to move forward in a batch is to re-write the system. RFID is just a tool, and assets are well managed, and ultimately the system.
- The receipt must be brushed (auto-identified by a reader) and the non-swipe card is considered not returned
- The inventory count was changed from regular to needs-based (system automatically detects anomalies and manually reviews)
- End-of-life process adds RFID write-off steps to avoid misreading of labels from the outside
Expect some resistance when the policy goes live. Operators may see it as one more step. Make the system as unobtrusive as possible: tapping a card should take under one second, reader antennas should cover the area so tags are detected on entry, and inventory data should upload automatically to the dashboard. The smoother the workflow, the less resistance to adoption.
Stage 4: Build up a data history and turn a stocktaking tool into a management tool
The value of the system began to appear when the entire RFID network was stable for more than six months. The problem of "conformity" was addressed in the first phase, at which point the focus shifted to "assets utilization" and "life cycle management".
Take a specific example. A set of tooling fixtures used to be scattered across four workshops. Whoever needed one had to look for it, spending 30 minutes on average. With RFID, positions are tracked in real time to within 3 meters, and staff can use a phone app to find the nearest fixture. Search time dropped from 30 minutes to 2 minutes. That saved well over CNY 100,000 in labor alone over a year.
The bottom layer of application is life expectancy. The system accumulates the frequency of use per asset, takes back records, maintains history, and, combined with a rated lifetime parameter given by the manufacturer, can calculate how long each asset may still last, pre-arranged for purchase and end-of-life. A sophisticated instrument manufacturer uses this mechanism to reduce spare parts stock by 25%, and the number of emergency purchases from 8 to 2 per month.
At this stage, Shanghai Selective Technologies typically advises clients to access RFID data into the BI viewer (i.e., the large video screen), and make visualized charts of asset locations, usage rates, idle rates, life-warning indicators. Management can see the state of the entire plant's assets in its office without having to ask a question on the ground.
The three most easily stepped pits before landing.
Finally, we'll give you a few typical cases of car flipping that we've seen.
First, the budget only counts hardware, not software and services. Many companies spend money on labels and readers, so back-office systems are free Excel templates, data can't be stored, analysed, and then reset. The RFID project has a reasonable budget allocation of about 40% hardware, 30% software, 30% implementation service, and this service is not saved.
Second, ignore the compatibility of existing systems. The RFID management platform is not an island, but it connects to ERP, MES, and financial systems. The interface development often involves more work than hardware deployment, and the IT sector must be involved in a very deep way in the preceding period, so that the RFID project does not become an information island.
Third, labels are cheap. Normal labels cannot read on metals, while outdoor showers and water fail. Metal resistance, high temperature tolerance (by the line oven), chemical erosion resistance, special labels, price differentials ranging from a few cents to a few dollars, but they bring about significant stability differences.
Summarize this phased path: First, you do asset classification and field surveys, then run the process in small-scale pilot runs, then you expand the synchronized system in batches, and then you upgrade the system from an inventory tool to a management tool with data deposition. Each stage gives you scope for retrospect and adjustment, from start-up to impact, 6-12 months of running, large businesses can pull up to 18 months, but don't come up at once and think about all the lines within one year, and the rhythm is too fast to roll over.

Ms. Lei
Mr. Pang