Insights / Inspection

A severity and urgency scale for facade defects that survives a handover

The value of an inspection report is not the list of findings. It is the ranking, and a ranking is only worth anything if the next inspector would produce the same one. That requires written criteria. This is a set that works.

Insights / A severity and urgency scale for facade defects that survives a handover

Published September 24, 2026 · 10 min read

Why "eleven findings" is not a result

Most facade inspection reports are inventories. Crack at grid B3, spalling at parapet east, staining below window 4-7. Eleven items, photographed, located, described. The building owner reads it and asks the only question that matters: what do I do first, and what can wait? The report does not say, or it says "recommend further investigation" against every item, which is the same thing.

The gap is a scale. Not a scoring model or an algorithm, just two agreed dimensions and written criteria for each level, so that two inspectors looking at the same photograph land on the same rank. Aviation maintenance has run on this for decades: a fault is either no-go, deferrable to a stated limit, or cosmetic, and the manual says which. The categories are boring. Their existence is what makes the operation work.

Two dimensions, not one

A single "priority" score mixes two different questions and answers neither. Severity asks how bad the defect is in itself: what it indicates about the structure or the envelope. Urgency asks how soon action is needed: what happens if nothing is done for a year. A hairline shrinkage crack is low severity and low urgency. A loose tile on the 9th floor over a pavement is low severity and very high urgency. A wide structural crack with no falling hazard is high severity and moderate urgency. Collapsing the two loses exactly the information the owner needs.

The scale below uses three severity bands, because more than three is false precision from a photograph, and a ten-point urgency score, because urgency has to be sortable across a whole building and three bands are not enough to sort eleven items.

Severity: three bands with written criteria

BandWhat it meansTypical indicators (surface)What it does not claim
HighDefect indicates a structural or envelope failure in progress, or a falling hazardCracks wider than about 1 mm running through units or across joints; spalling with exposed or corroding reinforcement; tile or render debonded and displaced; efflorescence or staining that maps to a crack patternDoes not confirm structural inadequacy. That is an engineer's finding, not a photograph's
MediumDeterioration that will become High if left through the next cycleCracks 0.3 to 1 mm; tile lift without displacement (hollow sound if tested); localized rust staining; sealant failure at joints with staining belowDoes not set a repair date on its own; see urgency
LowCosmetic or early-stage, monitor rather than actHairline cracks under 0.3 mm with no pattern; surface soiling; minor sealant weathering; isolated efflorescence with no crackDoes not mean ignore. It means record, photograph, compare next time

Two rules keep the bands honest. First, when in doubt, the higher band, because the cost of over-ranking is a specialist visit and the cost of under-ranking is a falling tile. Second, every High finding names the indicator that put it there, so a reviewer can disagree with the evidence rather than with the opinion.

Urgency: a ten-point scale driven by consequence and exposure

Urgency is severity multiplied by what is underneath. The same defect is more urgent over a school entrance than over a locked service yard, and more urgent at height than at ground level. The scale is anchored at four points and interpolated between them:

The exposure factor is the part inspectors most often leave implicit, and it is the part the owner cares about most. Writing it into the score is what turns a defect list into a plan.

Worked example: the same building, ranked

FindingSeverityExposureUrgencyAction
Tile displaced 15 mm, 9th floor, above main entranceLow (single tile)Public, height10Cordon today; remove tile; survey adjacent field
Crack 1.5 mm through three units, west elevation, rust stainHighCar park, low footfall7Engineer's inspection this quarter
Spalling with exposed rebar, parapet, rear service yardHighLocked yard6Repair in this quarter's package
Sealant failure with staining, windows 4-7, northMediumOccupied offices5Next maintenance cycle; check interior for ingress
Hairline map cracking, ground floor renderLowPublic, ground2Photograph; compare next survey

Note that the most urgent item is the least severe. A report sorted by severity would have put it fourth. A report sorted by this scale puts it first, which is where the owner's attention belongs on Monday morning.

What a photograph can and cannot support

Every band above is described in terms of what is visible at the surface, because this scale is designed for the first pass: photographs from the ground, from a drone, or from a smartphone at a window. The field guide to cracks, tile lift, spalling and efflorescence shows what each indicator looks like. The AI facade inspection in Airffic World applies a scale of exactly this shape to uploaded photographs and returns the ranked list.

What a photograph cannot support is a statement about the subsurface. Delamination behind an intact tile, corrosion behind sound render, and the structural adequacy of a cracked element are all invisible from outside. The scale handles this honestly: a High severity band is a reason to send a qualified engineer, not a substitute for one, and the report says so against every High item. In the aviation analogy, the first pass is the indication; the engineer is the inspection.

Why written criteria are the point

The scale above is not the only defensible one. Different climates, cladding systems and regulatory regimes will shift the thresholds. What does not change is the requirement that the criteria be written, so that the ranking is a property of the building and not of the inspector. When the inspector changes, the ranking should not. When the owner disputes a rank, the argument should be about whether the crack is really 1.5 mm, not about whether the inspector is being cautious.

That is the airline habit again: move the judgment out of the person and into the document, so the person can be replaced and the document can be improved. For how the same move plays out across a whole inspection program, see facade inspection without scaffolding; for where it sits in a renovation plan, the guide to surveying a building before large-scale repair covers the sequence.

Tsuyoshi Kiyomine came to AI from airline operations control and founded KIYOMINE in Tokyo. He builds Airffic World, a drone operations, AI inspection and rostering platform, and writes here about designing operations for work that moves.