Straight answer: the paper register survives because it is the exception handler for a machine that fails. Retire the failures and the register retires itself. Cut it on day one without doing that, and it will be back inside a month.
We see this on almost every site. There is a biometric machine at the gate, and there is also a register in the supervisor's drawer. Everyone treats the drawer as the truth.
Why the paper survives
The machine rejects some people. Worn fingerprints on manual workers. Each rejection needs a manual entry, and manual entries need somewhere to live.
The machine does not cover every door. Anyone using the side gate needs recording by hand.
Exceptions have nowhere else to go. Half-day approvals, site visits, someone who forgot to punch out. If the system has no path for these, paper absorbs them.
Trust, once broken, is sticky. After one payroll dispute the machine got wrong, supervisors keep the register as insurance. That is a rational response, not obstruction.
Notice that only the first reason is about biometrics. The rest are process gaps that would defeat any technology.
The four-week sequence that works
Week 1 — Parallel, no changes. Run the new system alongside the register. Change nothing else. Each morning, compare the two records for the previous day and log every mismatch with its cause.
Week 2 — Fix causes, not symptoms. Mismatches will cluster. Poor camera angle at one gate. An entrance nobody instrumented. A shift rule encoded wrong. Fix the top three causes. Do not fix the individual entries — fix why they happened.
Week 3 — Exceptions move into the system. Every remaining category of manual entry gets a defined digital path: who raises it, who approves it, what reason is recorded. This is the step most rollouts skip, and it is the one that decides whether the paper dies.
Week 4 — Register becomes read-only. Keep writing it, but payroll is now computed from the electronic record. Compare at month end. If the two agree, the register has no job left.
Retire it in week five, having proved rather than assumed.
What makes the electronic record trustworthy enough to stand alone
Two things, in our experience.
Complete coverage. Every entrance people actually use is instrumented. One uncovered gate keeps the paper alive indefinitely.
Evidence attached to each record. When a mark is disputed, being able to look at the frame the recognition came from ends the conversation. That single property is what converts supervisors from register-keepers into system-users — they stop needing insurance.
Camera-based attendance gives you both properties naturally, since every camera is a potential check-in point and every record stores its frame. That is what we build; the sequence above works regardless of which system you choose.
The one question that predicts success
Ask your supervisors: "If the machine and the register disagree tomorrow, which one will you believe?"
Until the answer is "the machine," you are running a paper process with expensive hardware in front of it.