Methodology
How the Relic Score works
The Relic Score is the number of years before replacing a building becomes more rational than continuing to repair it. This page is the whole derivation — the standard, the inputs, the arithmetic, the thresholds, and what the number cannot tell you.
The standard
The model is the ISO 15686-1 factor method, the international standard for service life planning of buildings. It is deliberately not a machine-learned model. A learned model on this data would be unauditable, would encode whatever biases sat in its training set, and could not explain itself to a homeowner whose house it had just condemned. The factor method can be checked by hand.
A component’s estimated service life is its published reference service life, multiplied by seven factors covering quality, design, execution, indoor and outdoor environment, use, and maintenance. We apply the factors we can honestly observe from an address and hold the rest at neutral.
The seven factors, and which ones we actually use
| Factor | Range applied | Why |
|---|---|---|
| A · Quality of components | Held at 1.0 | We cannot observe product grade from an address, so we do not pretend to. Stating this as neutral is more honest than guessing. |
| B · Design level | Era-derived | Folded into the construction-era factor below. |
| C · Work execution | Era-derived | Also folded into era — build quality tracks period practice more than anything we can see remotely. |
| D · Indoor environment | Held at 1.0 | Unobservable without entering the building. |
| E · Outdoor environment | 0.8 – 1.1 | Exposure. Coastal salt, freeze-thaw cycling and wildfire risk are the dominant external accelerants. |
| F · In-use conditions | Held at 1.0 | Occupancy intensity is not knowable from an address. |
| G · Maintenance level | 0.75 – 1.15 | The largest owner-controlled lever in the entire model. |
Four of the seven sit at 1.0. That is not laziness — it is the honest position. Component grade, indoor environment and occupancy intensity are not visible from an address, and inventing values for them would manufacture false precision in a number people will make six-figure decisions against.
Construction era
Design level and workmanship track building period more reliably than anything else observable remotely, so factors B and C are collapsed into a single era multiplier.
| Period | Factor | Reasoning |
|---|---|---|
| Before 1940 | ×1.10 | Surviving pre-war stock is self-selecting — heavier framing, old-growth timber, thicker masonry. What was poorly built is already gone. |
| 1940 – 1979 | ×0.90 | Post-war volume building: lighter framing, thinner assemblies, and the era of asbestos, lead paint and aluminium branch wiring. |
| 1980 – 1999 | ×0.95 | Better codes, but polybutylene supply, barrier EIFS and early engineered sheet goods. |
| 2000 onward | ×1.05 | Modern energy and structural codes, engineered lumber, materially better moisture detailing. |
Reference service lives
Reference lives are drawn from published durability data — ISO 15686 parts 1 and 8, ASTM E917 lifecycle costing practice, and NAHB/HUD component life studies. Published ranges are wide, and we deliberately use mid-range rather than optimistic values.
| Component | Reference life | Note |
|---|---|---|
| Structure — wood frame | 100 yrs | Sound framing kept dry outlives its finishes many times over; failure is almost always moisture, not load. |
| Structure — masonry | 150 yrs | Mass masonry is the longest-lived common residential structure. Mortar is maintained; the wall is not replaced. |
| Structure — concrete | 120 yrs | Governed by reinforcement corrosion, so cover depth and chloride exposure matter more than the concrete itself. |
| Structure — steel | 120 yrs | Effectively indefinite if never wetted; carbon steel in a damp assembly is a different and much shorter story. |
| Foundation | 100 yrs | Rarely the binding constraint, but effectively unrepairable at scale — which is why it caps the building. |
| Roof — asphalt shingle | 25 yrs | The dominant US covering and the shortest-lived. Manufacturer warranties routinely exceed real-world service life. |
| Roof — metal | 50 yrs | Fastener and coating life usually govern, not the panel. |
| Roof — tile | 75 yrs | The tile outlives the underlayment beneath it, which is the actual replacement cycle. |
| Roof — flat membrane | 22 yrs | Ponding and thermal cycling; the least forgiving of poor detailing. |
| Cladding — vinyl | 30 yrs | UV embrittlement and impact damage; colour failure usually precedes function failure. |
| Cladding — wood | 40 yrs | Entirely dependent on maintenance — the widest spread of any component here. |
| Cladding — stucco | 50 yrs | Long-lived when drained; barrier assemblies without a drainage plane fail far sooner. |
| Cladding — brick | 100 yrs | Outlives most of the building. Ties and flashings are the real service items. |
| Cladding — fiber cement | 50 yrs | Stable and well-documented; paint cycles rather than replacement. |
| Windows / glazing | 30 yrs | Insulated-glass seal failure governs, not the frame. |
| Electrical service | 45 yrs | Replaced for capacity, safety and code long before conductors wear out. |
| Supply plumbing | 50 yrs | Highly material-dependent — copper, PEX and polybutylene behave nothing alike. |
| HVAC plant | 20 yrs | The shortest-lived major system. Expect two to three replacements per structural life. |
The economic crossover
Remaining life is not simply when the frame gives out. Buildings are abandoned for economic reasons long before they are structurally unsafe. Replacement is called rational only when both of these hold:
- The spend is large. Components worth 45% or more of a full rebuild fall due within a rolling ten-year window — the point at which owners reliably stop repairing and start replacing.
- There is not enough building left to pay it back. Fewer than 25 years of structural life remain to amortise that spend over.
Both conditions are necessary, and the second is what stops the model being stupid. A sound masonry shell with every system spent is the classic renovation candidate, not a teardown — you replace the systems and get another century. Only when the frame itself is nearly gone does that spend stop making sense. Structural life is a hard ceiling throughout: you cannot outlive your frame.
Confidence, and why it is never high
A desk estimate from declared attributes is capped at medium confidence. When attributes are assumed rather than stated — which is the default before you tell us the year built — it drops to low, and the published range widens accordingly. The range widens further with age, because old buildings diverge: two identical 1920s houses can be fifty years apart in remaining life depending entirely on whether water was kept out.
We never publish the point estimate without its range. If you only take one number away from this page, take the range.
Era hazards
Some flags are raised from construction period alone — asbestos before 1980, lead paint before 1978, aluminium branch wiring between roughly 1965 and 1973, polybutylene supply from 1978 to 1995, barrier EIFS in the 1990s. These are stated as likely for the era, never as observed. Nobody has tested anything at this property. They appear because they materially affect insurability and renovation cost, and because finding them late is expensive.
What would sharpen the estimate
- Year built. The single largest improvement. Everything else is secondary.
- Roof age and material. The most common near-term failure and the one that cascades.
- Date of the last systems renovation. Resets electrical, plumbing, HVAC and glazing clocks.
- Honest maintenance history. Worth more than any other input except vintage.
- An actual inspection. Nothing here substitutes for someone in the crawlspace.
What this number cannot tell you
It cannot see a cracked foundation, a leaking valley, knob-and-tube in a ceiling void, or a beautifully maintained hundred-year-old house that will outlast its neighbours by decades. It does not know local land values, which in many markets drive teardown economics far more than building condition does. It is not an appraisal, not an engineering opinion, and not a substitute for a survey.
It is a defensible starting estimate from public method and declared facts — and it shows its working, which is more than most numbers attached to a house can claim.