A paper UHF label lists at four to ten rupees in India. An on-metal tag runs sixty to ninety. Take six for a label bought in volume and stick it on a ₹200 carton: 3%. Put a ₹60 on-metal tag on a ₹50,000 asset: 0.1%.
Nobody argues about the second one. The first is the one that adds up.
The stack
Two dock doors, three handhelds, one printer that encodes the tags as it prints them, an edge box, and the integration with your ERP.
A four-port Impinj R700 lists at $1,499. Four circularly polarised antennas to go with it run about $200 each. At today’s rate that’s roughly ₹2.2 lakh for a door before mounts and cable, so I’ve put ₹2.5 lakh. A Zebra RFD40 is somewhere between ₹1 lakh and ₹1.9 lakh depending on how far you want it to reach. It’s a reader sled, not a complete handheld: it needs a phone or a handheld computer to run it, and that host isn’t in the price. A ZD621R printer-encoder is about ₹2.2 lakh. Those are US reseller prices converted at ₹94.4, and what you’ll pay here depends on duty and on who you buy from.
Then the line people forget. Integration is the work of making a read mean something to the system that runs your warehouse: label formats, and which door is receiving and which is dispatch. How a read becomes a goods receipt. I’ve put ₹8 lakh. That’s a guess, and it’s the biggest guess in the stack. Get a quote for it before you get a quote for anything else.
All of it comes to about ₹20 lakh, once.
Tags are the budget
Now the part that runs every year. Two hundred thousand cartons at ₹6 is ₹12 lakh. Add ₹2 lakh for software and support and you’re paying ₹14 lakh a year to keep it going.
Over five years the tags alone come to ₹60 lakh. That’s three times the hardware and the integration put together. So when someone tells you a reader is expensive, ask what the tag costs, because the tag is what you’ll keep buying.
It’s also the lever that moves payback fastest. At ₹4 a tag the payback on these numbers is nine months. At ₹8 it’s thirteen. At ₹40 it never pays back.
Where it comes back
I’ve put four lines on the other side.
Cycle counts. Twelve counts a year at 200 hours each, at ₹150 an hour, with RFID cutting the time by 90%. That’s ₹3.2 lakh. The 90% is conservative: Auburn University’s RFID Lab and GS1 US reported counts 96% shorter in Project Zipper, a pilot with brands and retailers.
Hours at receiving and dispatch. Eighteen hundred a year, ₹2.7 lakh.
Shrinkage. Stock worth ₹20 crore, 1.5% lost a year, half of it recovered. ₹15 lakh.
Chargebacks. A hundred a year at ₹20,000, and a pallet read at the door before the truck leaves stops 80% of them. ₹16 lakh.
The ₹150 is the same figure I used for the barcode-or-RFID demo. Payscale puts a Mumbai warehouse worker’s base pay at about ₹2.2 lakh a year, which is ₹105 an hour, and PF and overheads take it to 140 or 150.
Now the honest part. The first two lines I can source. The last two I can’t. The National Retail Security Survey put US retail shrink at 1.6% of sales in 2022, and that’s sales, not warehouse stock, so my 1.5% is a guess that happens to look like theirs. The ₹20,000 a chargeback is a guess with nothing behind it at all. Together those two lines are ₹31 lakh of the ₹37 lakh a year the model says you save. Take them out and the tags and software cost more than the counts and the receiving hours save. It never pays back.
With them in, payback is about eleven months. The savings also arrive flat from the first month, with nothing discounted and no ramp-up, which flatters it. Your chargeback ledger knows the real number. Mine doesn’t.
99% is a per-tag number
This is the sum a quote leaves out.
A vendor says 99% read rate. That sounds like a pass, and it’s a statement about one tag. A pallet only counts as read if every case on it reads. With fifty cases that’s 0.99 multiplied by itself fifty times, which is 0.605.
Four pallets in ten leave with a case unaccounted for.
For 95% of pallets to read complete you need 99.9% per tag. That’s a different conversation, and it’s the one to have before you sign. The pallet drawing in the calculator marks the expected misses evenly, and real misses don’t spread out like that. They cluster in the middle of the shrink-wrap and against metal.
What a pilot has to show
Real stock, not sample cartons. Real doors, with the forklifts that use them. A month, not an afternoon. Count every miss against what the ERP says left the building.
And ask for the per-tag rate as a number you can check, not an adjective. If you want the physics behind why it varies, Wake the tag and the range calculator cover it.
Put your own numbers in
The ROI and read-rate calculator has every figure above as a field you can change, with where each default came from underneath it. It names the biggest line over five years. It also has the read-rate sum, with a slider for the per-tag rate you’d need.
Start with the integration line. It’s the one I’m guessing at, and it’s the one that’ll move.