How do you handle replacements over time?
By treating tags as consumables with a measured failure rate rather than as a finished installation. A deployment with no replacement process degrades silently, and nobody notices until a proportion of it is dead and the complaints have already stopped.
Measure your own rate
Count replacements per year across the deployment and record why each one failed: damaged, covered, painted over, removed with a fixture, never worked.
Those categories point at different fixes. A high damage rate is a specification problem, and a high covered rate is a placement problem.
A supplier's failure rate describes tags leaving a factory. Yours describes tags on your walls, and only the second predicts your budget.
Two years of your own numbers is what turns replacement from a surprise into a line.
The replacement inherits the identity
A replacement tag must resolve to the same record as the one it replaces, or the history is broken and the object appears new.
Where tags hold a written address, that means writing the same address. Where identity comes from the chip's own identifier, it means mapping the new one to the existing record.
Never reuse an identifier for a different object, which is how two things end up sharing a past.
And record the replacement with a date, since a tag replaced twice in a year is telling you something the summary figure hides.
Finding the dead ones
This is the hard part. A dead tag is silent, and the people who met it have gone.
Where tags point at a server you control, a location that stops producing arrivals is a candidate, and comparing locations against each other surfaces it.
Where they do not, the only route is physical inspection on a schedule, which is a real cost to budget.
Build a check into work that already happens: whoever changes the signage taps the tag while they are there.
Keeping stock
Keep a reserve from the same batch as the original run, because a top up ordered later is frequently a different product with the same name.
Verify the reserve when the main batch is verified, so replacements are not the moment you discover a dead blank.
Store it properly: cool, dry, flat, with nothing on top, and labelled with the batch and date.
And record where each tag was fitted, since the person replacing one in two years will not be the person who chose the position.
Frequently asked questions
How many tags will I replace each year?
Measure it rather than estimate it, and record why each one failed. The reasons point at different fixes.
Can a replacement have a new identifier?
It can be a new tag, and it must resolve to the same record. A new identity makes the object appear to have no history.
How do I find failed tags?
Comparing arrivals per location surfaces candidates where you control the destination. Otherwise it is physical inspection on a schedule.
Should I keep spare tags?
Yes, from the same batch, verified with the main run and stored properly. A later top up is often a different product.