Severity rating
What is a severity rating: severity rating explained
A severity rating is a scored judgement of how serious a defect is, placed on a defined scale so the same fault, recorded by different people on different days, lands in the same band and triggers the same response.
What does a severity rating measure?
Consequence, and how fast that consequence arrives. Severity asks what happens if nothing is done. It does not ask how likely the failure is, and keeping those two apart is the whole reason the rating is useful.
Likelihood belongs in risk scoring and in the risk assessment that sits around the inspection. Collapse severity and likelihood into one number too early and you lose the ability to explain the number afterwards. A supervisor reading "4" cannot tell whether that means minor damage that is certain to worsen or catastrophic damage that is unlikely, and those two findings need completely different responses.
What a defined severity scale looks like
A scale that survives contact with a real workshop defines each band by a required action and a deadline, not by an adjective. "Moderate" means nothing. "Rectify within seven days or take the asset out of service" means something a fitter, a planner and an auditor all read the same way.
- Band 1, immediate. The asset comes out of service now. The finding is reported verbally before the paperwork.
- Band 2, timebound. Safe to operate under stated limits, must be put right inside a named window, with the deadline written on the record.
- Band 3, monitor. Recorded, measured, rechecked at a named date. No work raised yet.
- Band 4, advisory. Noted for the owner. No action required, no clock running.
That structure is not invented. UK lifting equipment law works the same way: a competent person who finds a defect that is or could become a danger must say so immediately and confirm it in the report, and where a defect has to be put right inside a timescale, that timescale goes in the report itself (HSE, Thorough examination of lifting equipment, INDG422). The severity band and the response are the same statement.
Severity rating explained: a worked application
A counterbalance forklift comes in for its scheduled examination and three findings come out of it. A hydraulic hose is weeping at the fitting with no measurable pressure loss, so it goes to band 3 with a recheck in 30 days. The load plate is faded to the point where the rated capacity cannot be read, so it goes to band 2 with a seven day deadline. A lift chain measures past the manufacturer's stated withdrawal limit, so it goes to band 1 and the truck is keyed out before the inspector leaves the yard.
Nobody in that chain has to interpret the word "serious". Each rating carries an action and a clock, and the escalation path for band 1 is written down before anyone needs it.
Why severity ratings drift
Because they are applied by people, and people calibrate to their own recent history. An inspector who has spent a month on a badly maintained fleet quietly raises their bar. One who has been on new equipment lowers it. Neither notices.
The size of that effect is measurable. The US Federal Highway Administration studied 49 practising bridge inspectors from 25 state agencies across seven test bridges and found that only 68 percent of condition ratings fell within one point of the average, and 95 percent within two points, on a scale of 0 to 9 (FHWA, Reliability of Visual Inspection for Highway Bridges, FHWA-RD-01-020, 2001). These were trained, qualified professionals rating the same physical structures.
Drift is worse than noise because it is directional. If band 2 slowly becomes the place borderline findings go, your open-defect backlog looks stable while the fleet degrades underneath it.
How to calibrate a severity scale
- Anchor every band with photographs. Two or three reference images per band per defect type beat any amount of written definition.
- Double-rate a sample. Have a second person rate ten completed jobs a month blind, and track how often the two disagree by more than one band.
- Watch the distribution per inspector. If one person never issues a band 1 and another issues them weekly, the scale is not the problem.
- Retrain on a cycle. The same US bridge programme requires team leaders to complete 18 hours of approved refresher training in each 60 month period (23 CFR 650.309). Calibration decays whether or not you schedule for it.
- Kill the middle band if it swallows everything. A band that holds 70 percent of findings is a shrug with a number on it.
Severity rating, condition grading and claim severity
These three get mixed up constantly, and they answer different questions. A severity rating scores one defect. Condition grading scores a whole asset, usually against a commercial standard. Claim severity is an insurance measure of what a claim costs, which has no fixed relationship to how dangerous the underlying defect was.
Whichever scale you use, the rating only carries weight if the inspection report shows what it was applied to and who applied it. A number without its evidence is an opinion with a decimal point.
