A robot with four sensors on it can still be blind to the one failure that actually takes it down. Coverage percentage counts sensors. It doesn't ask whether any of them can see the specific failure mode heading for that asset. iFactory, an on-premise AI platform for predictive maintenance and OEE analytics, maps every critical asset's dominant failure mode against what its current instrumentation can actually detect — not just whether a sensor exists. Map your own failure-mode coverage.
Fully Sensored Doesn't Mean Fully Covered
A sensor on the asset isn't the same as a sensor that can see the failure mode that's actually going to happen.
"80% Covered" Was Never The Right Question
A coverage percentage counts sensors installed against assets that have them. It says nothing about whether the sensor installed can see the failure actually coming.
- Vibration sensors → catch bearing wear, shaft misalignment, gear defects.
- Winding insulation breakdown or slow current drift → different physics, different signal, vibration can't see it.
An asset with a vibration sensor and a motor about to fail electrically isn't 80% covered. It's 0% covered for the failure that's actually coming, with a sensor report that says otherwise.
No Single Sensor Type Sees Every Failure Mode
Four sensing modalities exist for a reason — each one opens a different diagnostic window, and none of them covers all the others.
| Sensor type | What it sees | What it misses |
|---|---|---|
| Vibration | Bearing wear, misalignment, gear defects, looseness | Electrical faults, thermal degradation |
| Thermal | Overheating, insulation stress, friction buildup | Early-stage mechanical wear |
| Current | Motor rotor faults, winding issues, load variation | Pure mechanical bearing defects |
| Pressure | Seal degradation, flow restriction, cavitation | Rotating-element faults on dry assets |
A credible coverage map fuses more than one modality on any asset with more than one plausible failure mode — which, on a stamping press or a welding robot, is nearly every asset on the floor.
Which Of Your Critical Assets Has A Blind Spot Like This?
Send your current sensor inventory — the team will cross-check it against each asset's dominant failure mode, not just whether a sensor exists.
Assets That Have Never Failed Look Covered. They Aren't Tested.
A sensor nobody has seen catch a real failure is a hypothesis, not a safety net.
- No failure history → no proof the alert threshold was set correctly, only a guess.
- Green on the dashboard → means the sensor is present, not that it works for this asset's actual failure mode.
That combination is a poor candidate for "first deployment confidence," whatever the dashboard shows. The sensor exists. Whether it catches anything useful has simply never been tested.
Green On The Dashboard. Blind To What Mattered.
A body-shop welding robot shows full sensor coverage on the plant's instrumentation dashboard — vibration sensors on the arm joints, right on schedule for the PdM rollout. The robot fails anyway, mid-shift, from a servo gearbox losing backlash tolerance gradually over weeks. Nothing in the vibration signature caught it; backlash drift shows up in torque feedback, a modality nobody installed because the dashboard already said "covered." The coverage report was accurate and useless at the same time — accurate about sensors present, useless about the failure that actually happened.
Four Questions Per Asset, Not One Percentage For The Plant
- What is this asset's dominant failure mode? Not "what could theoretically fail" — what has actually failed here, or on equipment like it, before.
- Which sensing modality actually detects that mode? Match the physics, not the catalog — a vibration sensor doesn't help against a winding fault.
- Is that modality installed, or just a different one? "Sensored" and "sensored for the right thing" are not the same claim.
- How early does the installed sensor catch it? A sensor that sees the fault two days before failure barely beats a run-to-failure strategy; two to six weeks is what turns maintenance into a scheduled event instead of a scramble.
Rank By Consequence Times Detectability, Not Alphabetically
Not every gap deserves the same budget line, and not every asset deserves a sensor on day one.
- Operations ranks by downtime cost → doesn't account for whether the failure is even detectable.
- Maintenance ranks by repair difficulty → doesn't account for how much warning a sensor can give.
- IT ranks by data volume already flowing → rewards assets that are easy to monitor, not assets that matter.
The right starting list ranks by failure consequence and frequency, filtered by whether the dominant failure mode is detectable at a useful lead time. A high-consequence asset with an undetectable failure mode isn't a sensor gap to close this quarter — it's a different problem, and chasing it first burns budget the detectable gaps would have used better.
"We Can't Sensor Every Failure Mode" Isn't A Reason To Skip This Exercise
The point of a failure-mode map isn't universal coverage — it's knowing exactly where the blind spots are before one of them costs a shift.
- Some failure modes stay undetectable → that's fine, as long as it's a documented decision, not an unknown gap.
- Budget always runs out before the asset list does → ranking by consequence and detectability means it runs out on the right assets last.
The Map Is Failure Modes Against Sensors, Not Sensors Against Assets
iFactory starts from each critical asset's known or likely dominant failure mode, then checks what the current instrumentation can actually see — fusing vibration, thermal, current, and pressure data where more than one failure mode is plausible, which is most of the time.
- Coverage scored per failure mode, not per asset — a station can be covered for one mode and blind to another.
- Gaps ranked by consequence and detectability, so budget goes to the fixable high-impact gaps first.
- New sensors fused with existing instrumentation rather than replacing it outright.
- Alert lead time tracked per failure mode, so "covered" means weeks of warning, not a same-day alert nobody can act on.
What A Reliability Lead Should Check Before Trusting A Coverage Report
A coverage report earns trust when it names the failure mode each sensor actually catches, not just the sensor count per asset.
What Reliability Teams Ask About Sensor Coverage Gaps
Not on its own — it tells you sensors exist, not whether they can see the specific failure mode heading for that asset. Walk through the difference with your own asset list.
If the installed sensor type doesn't match the asset's dominant failure mode — vibration on a motor about to fail electrically, for instance — the sensor exists but the gap does too. See your own mismatches mapped out.
No — fuse modalities only where more than one failure mode is plausible, and rank by consequence and detectability rather than instrumenting everything identically. Ask support how prioritization is scored.
That's a different problem than a coverage gap — it means the asset isn't a good candidate for sensor-based PdM yet, whatever its consequence ranking. Talk through the alternatives with the team.
Weeks of lead time turns a failure into a scheduled maintenance window instead of a scramble — a same-day alert barely improves on finding out when it breaks. See real lead times for your asset types.
Map Your Own Failure-Mode Coverage Gaps
Book a session with the iFactory team and find out which of your critical assets are sensored for the wrong failure mode.
