How the data actually gets pulled from your elevator equipment
Otis places sensors directly on the controller, the motor drive, and the door operator assembly. These feed into their Connect platform, which pushes structured event logs to a cloud backend. The raw feed includes trip counts, floor call tallies, door cycle timing, energy draw per trip, and fault codes logged at roughly one-second granularity. That's the foundation. Everything else is built on top of this. The dashboard surfaces uptime percentages, traffic pattern trends, and comparative benchmarks. Most people stop there. The actual useful signal is in the door performance curves and the energy delta between trips. Those two numbers tell you whether your system is degrading or performing normally. Uptime percentages are useful but they don't capture short interruptions that matter to occupants.
The Otis Track How Datadriven Elevator Performance Analytics Are Revolutionizing Building Efficiency And Safety
This isn't just a branded monitoring screen. The analytics engine is built to detect patterns that dispatchers and building engineers routinely miss. It flags doors that are taking progressively longer to close without triggering any explicit fault code. A facility manager won't notice a half-second increase over three months. The algorithm correlates it with passenger volume, ambient temperature shifts, and door motor current draw. It surfaces the issue before it becomes a service call. Safety analytics take a different path. Instead of waiting for a hard fault, the system monitors vibration signatures from the hoistway and compares them against baseline profiles established during commissioning. If a counterweight guide shoe wears unevenly, the vibration signature changes before the elevator triggers a safety alarm or passengers feel any discomfort. You get a maintenance notification weeks ahead of a potential breakdown.
Reading the metrics correctly
Uptime is the headline number, but the calculation matters. Otis defines downtime as any event where the car is out of service for more than ten minutes due to a fault. Short interruptions from normal traffic management do not count. A building can show 99.2 percent uptime while passengers are still experiencing long waits during peak hours. Don't let the percentage fool you. Energy per trip is more actionable than total energy consumption. It removes the variable of building occupancy and lets you compare efficiency across different floors or different time periods. A properly tuned regenerative drive system can cut energy per trip by 30 to 40 percent compared to resistor-based braking, but only if your traffic pattern justifies the regeneration cycles. Low-traffic buildings don't see the full benefit. Door cycle consistency is probably the most underrated metric in the platform. Inconsistent door timing is usually the first sign of mechanical wear or a failing control board component. When doors start behaving erratically, the root cause is often the force sensor calibration or the door operator speed curve, not the obvious symptom of slow closure. I've spent entire diagnostic visits chasing door delays only to find a drift in the position encoder that the analytics had flagged weeks earlier.
Get the Full Details
How I deal with false positives on older retrofitted equipment
I ran into a real problem with a 1998 GenieSIGHT retrofitted system in a downtown high-rise. The analytics flagged a traction sheave wear issue based on vibration data, but the building's HVAC system was cycling on the same floor and creating false vibration readings. I pulled the HVAC runtime logs from the building management system and cross-referenced them with the flagged events. The correlation was clear. About 80 percent of the alerts occurred during HVAC cycle windows. The workaround was creating a time-of-day exclusion filter in the analytics platform to ignore data from those windows. It eliminated the noise and the remaining alerts turned out to be legitimate. I then recommended replacing the loose accelerometer mount on car three, which had been amplifying normal vibrations into false readings. Once the mount was secured, the filter wasn't needed anymore. The system worked correctly, but only after I removed the environmental interference.
Where this approach completely fails
Predictive analytics depend entirely on sensor quality and correct installation. If the accelerometer was mounted loosely during a retrofit, the data is useless regardless of how sophisticated the algorithms are. I've seen multiple jobs where vendors installed sensors on unsecured brackets, and the resulting vibration data looked like random noise. The platform generated hundreds of false maintenance alerts before anyone noticed the mounting hardware was the problem. Check your sensor installations annually. It takes twenty minutes and prevents months of noise. The system also assumes you're running connected equipment. Legacy controllers without communication modules won't feed data into the platform. Retrofitting them is expensive and sometimes impractical on older installations. In those cases, you're better off with manual inspection schedules and basic trip counters. Don't force a analytics solution onto hardware that wasn't designed for it. It will cost you money and produce nothing useful. Another limitation nobody advertises is that the analytics don't replace skilled maintenance technicians. The system tells you that something is deviating from normal. It doesn't tell you why. A door operator drawing excessive current could mean a worn belt, a binding guide, or a failing capacitor in the drive. The algorithm flags the anomaly. The technician diagnoses the root cause. There's no shortcut around that.
Getting the platform deployed
If you're managing a fleet of Otis equipment, the first step is confirming all controllers have active Connect subscriptions. The base monitoring package covers trip counts, basic fault logging, and uptime reporting. The advanced analytics layer with predictive maintenance features requires additional licensing per elevator. Pricing varies by region and contract terms. Work with your local Otis representative for actual figures. Don't assume the base package includes everything advertised on the marketing page. Data export is available through the platform, but granularity depends on your subscription tier. Basic tiers give you daily aggregates. Higher tiers unlock minute-level event logs and raw sensor data for custom analysis. If your maintenance team needs to do deeper investigation, verify export access is included before signing the contract. Losing access to raw data after deployment is a common complaint. The mobile app provides real-time alerts and basic status checks, which is useful for on-call technicians. The full analytical workspace is web-only. Don't expect to do serious data work from the phone. The interface is optimized for notifications, not investigation.
Things the documentation doesn't emphasize
The benchmarking feature compares your building's performance against a database of similar Otis installations. This is genuinely useful for identifying outliers, but the comparison groups can be misleading if your building has unusual traffic patterns. A hospital with emergency department traffic will look inefficient next to a Class A office building, even when both are running correctly. The system doesn't always account for these differences automatically. Use the benchmark as a rough guide, not a definitive judgment. The system also learns your building's baseline over time. The first 30 to 60 days after installation are essentially a calibration period. Alerts during this window are less reliable because the system hasn't established normal operating parameters yet. Don't panic if you see a spike in flagged events right after go-live. Give it a month to stabilize before making maintenance decisions based on the analytics. Data retention policies vary significantly by region. Some locations store raw event data for 12 months. Others keep it for only 90 days. If you need historical data for long-term trend analysis or warranty disputes, verify the retention period in your contract before committing. I've dealt with situations where six months of fault history disappeared because of an unexpected data purge policy change. Having that history available during a warranty claim made the difference between a covered repair and an expensive out-of-pocket replacement.
The analytics platform is a useful tool when you understand what it can and can't do. It catches problems early. It quantifies performance trends. It reduces emergency service calls. But it doesn't install itself correctly, it doesn't replace qualified technicians, and it doesn't work on equipment that can't communicate. Treat it as a decision-support system, not an autonomous maintenance solution. That mindset will save you time, money, and a lot of unnecessary service calls.
