The market context
The world installed more elevators
than manual inspection can keep up with.
Three curves crossed at the same time: the installed base got older, the workforce got smaller, and the standard of proof that owners, regulators and insurers demand got higher. Calendar-based maintenance was built for none of those conditions.
01 · The installed base
Aging isn’t a future problem. It’s already here.
Elevators are 20-to-30-year assets installed in a 40-year building boom. The units that went in during the peak years are all reaching the same part of their wear curve at the same time.
Sources: China market regulation registration data (2025); European Lift Association SNEL White Paper (2020); NEII and US Bureau of Labor Statistics Occupational Outlook Handbook (2025); London Fire Brigade open data (2026).
02 · The labor gap
You cannot hire your way out of this.
An elevator technician takes years to train. The installed base does not wait for them. Every market that has published the numbers shows the same shape: units per technician rising, and the experienced half of the workforce approaching retirement.
The scarce resource is not the technician’s hands. It is the technician’s attention — and attention is the one thing a measurement system can allocate.
Routing is guesswork
Today’s route is set by the calendar and by whoever complained loudest. Neither correlates with which elevator is actually degrading.
Experience does not transfer
The senior technician who can hear a failing bearing is retiring. Without a recorded signature, that judgment leaves the company with him.
The second trip eats the margin
Arriving without knowing the fault means diagnosing on site, ordering the part, and coming back. The return visit is pure cost.
03 · The physics
Mechanical failure develops slowly. Discovery feels sudden.
A bearing does not fail all at once. Mechanical degradation develops over time, and the change can appear in vibration and acoustic signals before it becomes obvious to a person or is logged as a fault code.
Illustrative degradation curve based on the three-phase escalation pattern observed in the traction-machine case. Exact timing varies by component and duty cycle.
Different detection methods become useful at different points in the failure curve. The earlier a developing anomaly becomes visible, the more time there is to inspect and respond before it escalates.
04 · Why this is possible now
Condition monitoring is old. Doing it on 11 million elevators is new.
Vibration analysis has been standard practice in aerospace and heavy industry for decades — on assets worth millions, with dedicated engineers reading the spectra. Four things changed that brought the same method within reach of a residential elevator.
MEMS sensors got cheap
A 6-axis accelerometer with the resolution to resolve bearing frequencies now costs a fraction of a single callback visit.
Connectivity stopped needing a project
4G modules removed the building-IT negotiation. No VLAN, no firewall exception, no wiring closet.
Models learn per-unit baselines
The breakthrough is not classifying a fault — it is learning what normal sounds like for one specific 18-year-old elevator, so it is judged against itself.
Proof became commercially valuable
Owners, regulators and underwriters now all ask the same question — can you show me? Reliable data helps every party to the contract make a better decision.
05 · The cost of doing nothing
The status quo has a cost. It is simply harder to see.
Every one of these is already being paid for somewhere in the contract — usually by whoever is least able to see it coming.
| What happens | Who pays today | What it costs | Detectable in advance? |
|---|---|---|---|
| Entrapment call-out | Maintenance company | Emergency crew, out-of-hours rate, fire brigade attendance, reputational damage with residents | Yes — timing varies by fault and operating conditions |
| Unplanned shutdown of a traction machine | Building owner | Expedited parts, unplanned crane or hoist access, days of a core asset offline | Yes — timing varies by fault and operating conditions |
| Repeat visit for the same fault | Maintenance company | A second full crew-day, and a customer who now doubts the first repair | Yes — verifiable on site |
| Disputed maintenance quality | Everyone | Contract renegotiation, withheld payment or disputes without a shared traceable record | Only with before/after data |
| Premiums priced on averages, not on this building | Owners who maintain well | Good fleets subsidize bad ones, because nobody can tell them apart | Only with per-unit risk data |
| Loss of a retiring technician’s judgment | Maintenance company | Years of tacit knowledge that was never written down in any form a successor can use | Not without a recorded signature |