An auditor asking about harness compliance is rarely asking whether the webbing is sound. The question underneath is what happened between the checks: which hooks were open during work, at which anchor height, on which crew. Sampling inspection cannot answer it; continuous records can.
One distinction first. This page is about proving use. Proving the equipment condition is a different exercise with its own checklist — see the harness inspection checklist. Audits that conflate the two usually produce a folder of inspection sheets that say nothing about behaviour.
What sampling can and cannot prove
A sampled compliance check — a supervisor walks a floor, looks at five crews, signs a sheet — proves three things about one moment: the workers observed were connected when observed, the equipment looked sound, and someone was supervising. That is worth having. What it cannot prove is anything about the other hours of the shift, or about the crews not sampled, or about the state of a connection at the moment a worker changed position. The gap between checks is not a small residue of the audit risk; for tie-off behaviour it is essentially the whole audit risk, because disconnection happens quickly, deliberately and out of sight.
Regulation points the same way. Pre-use inspection of personal fall protection systems before each workshift is required of the equipment (29 CFR 1910.140(c)(18)); the connected-at-the-moment-of-work behaviour has no inspection clause to lean on, because it cannot be inspected after the fact. It can only be observed live or recorded as it happens.
What continuous compliance records look like
Continuous monitoring turns each state change into a timestamped record tied to a specific harness and shift. The event set that matters for fall protection:
1. worn at height or not worn;
2. hook open or closed, per hook on a twin-tail lanyard;
3. attachment below the wearer — the foot-level anchor that looks fine from the
ground;
4. alarm raised, and whether it was acknowledged and corrected.
A multi-state harness monitor (SBM3000 series) generates these on the unit itself, so the record does not depend on a network being available at the moment of the event; alarms sound locally at once and the events queue for upload. It takes no part in load bearing and does not act on the system — the protection stays in the harness, lanyard and anchor, and the monitor's job is the evidence.
Building the evidence chain
Records only become an evidence chain when three things are fixed in advance. First, what counts as an event: an open hook during work at height is an event; an open hook at ground level during rigging may not be — the configuration has to say which. Second, who reviews and how often: a shift-end glance for anomalies, a weekly review that groups events by crew and work front, a period review that feeds decisions. Third, where the record lives and for how long, in a form an auditor can read without a translator.
What the chain is worth shows up in the decisions it feeds. Event patterns by work front point to where the next engineered anchor should go; repeat open-hook events on one crew point to a training gap rather than an attitude problem; anchor-height events confirm or retire the "low anchorage is rare" assumption. A sampled walk-round produces none of these, because it never sees the moments the decisions are about.
Sampling and monitoring are not rivals. Sampling stays for the equipment condition it is good at; monitoring covers the use-behaviour interval sampling is blind to. An audit file with both reads as a programme rather than a spot-check.
Questions buyers ask
Does continuous monitoring mean watching workers?
The record is about equipment states, not about people's movements: which hook was
open, at what height, when. Most programmes share event records with crews rather than
individual dossiers, which is also what gets the cooperation the data depends on.
What if the site has no network in the work zone?
The alarm and the record are generated on the harness unit itself; events queue on the
device and upload when the link returns. A dead zone produces a later record, not a
missing one.
Can this replace the inspection programme?
No. Inspection covers deterioration the electronics cannot see — stitch lines,
buckle wear, label legibility. Monitoring covers the connection states between
inspections. The two together are the audit file.
About Yunku Technology
Yunku Technology builds the SBM3000 multi-state smart safety harness: wear, hook and anchor-height detection with edge alarm logic and timestamped event records, described on the multi-state safety harness page. For the definition and capability boundary of the category, see what is a smart safety harness.