How to increase your soiled-in RFID accuracy
The importance of 'soiled-in' scanning
Every laundry that has implemented, or attempted to implement, RFID has had issues with accuracy. The sheer volume of items passing through seems to suggest 100% is impossible. But the soiled-in scanning is hands-down the most important aspect — a missed read receiving linen affects the data very differently to a missed read shipping it.
Example: Obie's Linen Services
Take a fictional laundry, Obie's Linen Services, servicing Alice's Restaurant. On Monday, 100 tablecloths go out and through the trolley scanner — but only 98 are scanned.
From the laundry's perspective: it creates uncertainty. Are the missing two lost?
From Alice's perspective: no issue — she isn't held accountable for those two sheets.
Now say Alice pushes that trolley of 100 off a cliff. At month end Obie can still confidently account for 98 sheets and fairly charge for the loss.
Now flip it: inaccuracy on the soiled-in side
Obie's sends 50 serviettes, scanned out correctly. After a function Alice sends all 50 back — but only 48 scan in. "Looks like we never got back 2." "We definitely sent them all back…"
Inaccurate soiled reads create false positives; inaccurate shipping reads create false negatives. False negatives are silent. A false positive leaves the laundry believing linen is still at the customer — creating conflict, or worse, erroneous charges.
The Swiss Cheese Receive
So how do we get accuracy in soiled-in scanning? We borrow from our Swiss dairy products. In risk analysis, the Swiss Cheese Model shows how holes (failures) in multiple layers can align — but the more layers you add, the lower the chance the holes line up across all of them. Our holes are missed reads. Multiple read points, each well below 100%, quickly add up to a number very close to it.
The maths
Picture one conveyor carrying all soiled linen, with a single reader catching 95% of tags. Total accuracy: 95% — for every million items, 50,000 missed.
Add a second reader on the CBW chute at 95%. The chance a sheet is missed by both is 0.05 × 0.05 = 0.0025 → 99.9975%. Still 2,500 missed per million.
Add a third behind the dryers at 90%: 0.05 × 0.05 × 0.1 = 0.00025 → 99.99975%, or 250 per million.
Finally, make the dispatch cabinet the last line of defence — anything shipped that wasn't already picked up counts as a soil-in. At 98%: 0.05 × 0.05 × 0.1 × 0.02 = 0.000005 → 99.999995%, just 5 sheets per million. A number you can waive to the customer in advance.
Audit, audit, audit
It's not enough to tell your customer your soiled-in process is accurate — you have to prove it. Auditing is mostly operational, not technological. Put a pretend customer on site, ship linen to them, return it, and process it as normal. You know with certainty you sent it all back — so what does the RFID report show? If the system's healthy, 100% comes off the account. Log it, then repeat daily.
That does two things: it flags problems fast, and it gives you something tangible to show customers. "Here's six months of daily audits showing the accuracy of our soiled-in process" beats "Trust me, bro." With the Swiss Cheese Receive and regular audits, you can assure your customer — and yourself — that nothing is falling through the cracks. Trust in the system is paramount if you want real value from your RFID infrastructure.
