POM365
Methodology & standards

How the POM Score is built

A transparent, 0–100 measure of a home's condition and health — grounded in the same standards appraisers, inspectors, and lenders already use. This page is the citeable reference for agents, appraisers, inspectors, and investors. Save it (⌘/Ctrl-S) or print to PDF.

In one sentence

The POM Score is a component-weighted effective-age / condition index from 0 to 100: it estimates how a home is actually holding up — its structure, systems, updates, documentation, and resilience — by scoring each part against its known useful life, and it improves only when real facts improve.

Why it's defensible. Every input traces to an established standard — the appraiser's cost-approach "effective age" depreciation, the home-inspection industry's component life-expectancy tables, and the mortgage industry's UAD condition ratings. Nothing is invented; the weights are the only tunable, and they're calibrated against real outcomes over time.

1 · The standards it's built on

DisciplineStandard we useWhat it grounds
AppraisalCost approach — effective age ÷ economic life depreciation (Appraisal Institute / Marshall & Swift)1How condition & age convert to a score; how it depreciates over time; the 0–100 ceiling.
InspectionNAHB Study of Life Expectancy of Home Components2 + InterNACHI Estimated Life Expectancy Chart3The useful-life years for roof, HVAC, water heater, etc.
LendingFannie Mae / Freddie Mac UAD condition C1–C6 & quality Q1–Q64The band labels — defensible to a lender or appraiser.
Cost & ROIRSMeans (Gordian) construction cost data5 + Cost vs. Value (Zonda)6Renovation cost ranges and recouped-value %.

2 · The five pillars

The score is a weighted blend of five pillars. Weights favor what actually costs a homeowner money and risk — the structure overhead and the systems inside — while still rewarding updates, transparency, and a resilient location.

PillarWeightWhat it measures (and its standard)
Structure25%Roof, foundation, walls, windows — scored on remaining useful life (NAHB/InterNACHI).
Systems20%HVAC, water heater, electrical, plumbing — remaining useful life.
Interior & Updates15%Kitchen/bath recency & build quality (UAD Q1–Q6).
Documentation15%Permits, disclosures, inspection on file, clean title.
Resilience & Location25%Hazard exposure (FEMA NRI), insurability, neighborhood trend.

3 · The formula

Each component scores on how much of its useful life remains (new ≈ 100, end-of-life ≈ 10). Each pillar blends what we measured with how confident we are in it:

Pillar score  =  confidence × measured + (1 − confidence) × conservative prior
POM Score  =  round( min( Σ (weightᵢ × pillarᵢ),  confidence ceiling,  quality ceiling ) ), clamped 0–100
…then any confirmed safety/structural defect caps the score below 40 — mirroring the appraiser's "subject-to" (C6) rule.

Confidence rises with the source of each fact: public record → owner-stated → permit-matched → documented → independently verified. Better evidence both moves the score toward the truth and lifts the ceiling that gates the top "Pristine" band — which is what makes proving facts worthwhile.

4 · The tiers (mapped to appraisal condition)

POMTierPlain meaning≈ UAD
90–100PristineNew or fully renewed, top build quality, verified, low risk. Rare.C1
80–89ExcellentWell kept, updated, strongly documented.C2
70–79HealthyNormal wear for its age, well maintained.C2–C3
55–69FairDeferred maintenance creeping in, systems aging.C3–C4
40–54Needs AttentionSignificant deferred maintenance.C4–C5
0–39At RiskMajor systems failing or a safety issue.C5–C6

Thresholds are a display layer over the same 0–100 number and are tuned against real outcomes; the underlying math is unchanged.

5 · Unknown data is a conservative prior, never a zero

When a component's age is unknown, we do not assume the worst. Following standard appraisal practice (absent contrary evidence, a component is presumed to date to the home), we seed its age from year built and mark it lower-confidence. The home shows a "Limited data — confirm to sharpen" note rather than a punitive label. Confirming the real age moves the number toward the truth and raises confidence.

6 · Depreciation over time is automatic

Because each component scores on remaining useful life, the score drifts down naturally as a home ages — exactly like straight-line appraisal depreciation — as long as it is recomputed against today's date. POM recomputes on a monthly cadence, so the number reflects aging. Maintenance and renovations lower a component's effective age (raising the score); time raises it (lowering the score).

7 · Why the score can't exceed 100 — and why per-project lifts don't simply add

This is the question a sharp reviewer asks, so here is the precise answer:

The POM Score is never an additive point tally. It is always a fresh recompute from the home's current facts, clamped 0–100. A home at effective age 0 (as-new) is the best a condition score can be — you cannot be "younger than new," so 100 is a true ceiling, mirroring the appraiser's 0% depreciation / C1.

So why don't the "+N" project lifts add up to more than 100? Each project's "+N POM" is a marginal estimate from the home's current state — the lift if you did only that one thing next. Because of diminishing returns near the top, every successive improvement adds less: a project that shows "+5" at a score of 80 might only add "+2" once the home is at 92. The lifts therefore shrink as you approach 100 and cannot overshoot it.

That's why the plan panel shows a recomputed projected total ("your plan takes you 70 → 81"), not the sum of the individual "+N"s. The individual numbers are labelled directional estimates; the plan total is the honest, clamped recompute. And because reaching 100 requires near-new condition across all five pillars plus full verification and low risk simultaneously, the ceiling almost never binds in practice. Where two homes both top out, the POM Verified badge and the dollar value continue to differentiate them — so motivation doesn't stop at the cap.

8 · Renovation cost — scaled to the home, from contractor cost data

Costs are never a flat number. They follow the contractor/appraiser method — unit cost × quantity × regional factor — where quantity derives from the home's size:

ProjectDriver (quantity)Unit cost basis5
RoofRoof squares = footprint × pitch (≈1.1–1.4); 1 sq = 100 sq ft≈ $350–$900/sq (3-tab) · $500–$1,500+/sq (architectural)
HVACTons ≈ sq ft ÷ 500–600 (climate-adjusted)≈ $1,900–$2,300 / ton installed
Kitchen / bathTier (minor / mid / upscale) × sizeCost vs. Value tier ranges6

A ~1,700 sq ft single-story has ~18–22 roof squares; a 10,000 sq ft home is several multiples — so its roof cost is proportionally higher. Production source: RSMeans (the construction industry's standard cost database) and Zonda Cost vs. Value, regionalized by ZIP. Confidence: directional (±~20–30%) on a unit-cost model; tighter with a licensed RSMeans / Xactimate feed. Always shown as a range, labelled an estimate — not a bid.

9 · Value impact — and how it layers on the AVM

Two different numbers, kept separate: the condition (POM Score) and the dollar value. Value is built in layers, most-rigorous last:

LayerMethodConfidence
1. AVM baselineATTOM/CoreLogic estimate from public data±~7% (off-market)7
2. Condition adjustmentPOM shifts value within the AVM band (capped); updated facts (sq ft, beds, baths, quality) re-run the AVMdirectional
3. Project recoupCost vs. Value % recouped per project (regional) — never the full costdirectional
4. Agent CMA / appraisalA licensed professional's opinion — overrides the AVM for material changeshighest
Why "added value ≠ cost." The market rewards improvements by recoup %, not receipts — and over-improvement (a luxury upgrade beyond the home's tier — e.g., a smart fridge in a modest kitchen) recoups little to nothing (the appraisal principle of conformity / diminishing returns). So POM credits discretionary upgrades conservatively and escalates anything material to an agent CMA.

10 · Confidence & the path to higher

Methodology — defensible today

~88%. Built entirely on established appraisal, inspection, and lending standards (above). An appraiser or inspector can follow every step.

Calibration — improves with data

~65% and rising. The weights and value magnitudes get validated against real sale prices and appraiser ratings as volume grows.

To raise it: (1) calibrate weights against actual sales & appraiser C-ratings (the data flywheel); (2) license RSMeans/Xactimate for cost precision; (3) a formal review by licensed appraisers & inspectors ("validated by…"); (4) track prediction error and recalibrate quarterly. The model is versioned (snapshots are immutable) so every historical score is reproducible and auditable.

11 · Worked example — 5083 Bird Nest Trail

2018 build · roof 2020 · HVAC 2018 · tankless water heater 2018 · PEX plumbing · well maintained · Tallahassee, FL (moderate wind risk).

PillarScore× weight
Structure7819.5
Systems6613.2
Interior & Updates598.8
Documentation7010.5
Resilience7418.5
POM Score70 · Healthy

A well-kept 8-year-old home with young-but-original systems reads Healthy (70) — not "Fair." (An earlier build mislabeled it 62 by scoring unknown facts as zero; §5 fixes that.)

Sources

1. Appraisal Institute, The Appraisal of Real Estate — cost approach & effective age/economic-life depreciation; CoreLogic/Marshall & Swift cost service. appraisalinstitute.org
2. National Association of Home Builders, Study of Life Expectancy of Home Components. nahb.org
3. InterNACHI, Standard Estimated Life Expectancy Chart for Homes. nachi.org/life-expectancy
4. Fannie Mae Selling Guide B4-1.3-06, Property Condition & Quality of Construction (UAD C1–C6 / Q1–Q6). selling-guide.fanniemae.com
5. RSMeans (Gordian) construction cost data; Verisk Xactimate (repair cost). rsmeans.com
6. Zonda, Cost vs. Value Report (annual, regional recoup %). costvsvalue.com
7. Zillow Zestimate published median error (~7% off-market) — AVM accuracy benchmark. zillow.com/z/zestimate/
8. FEMA National Risk Index (hazard) · LexisNexis C.L.U.E. (insurance loss history).
POM365 — POM Score methodology. Model v1.1 · reference tables stored as versioned config; this page served at /methodology. Estimates are informational and are not an appraisal, inspection, bid, or guarantee of value; actual costs and values vary by vendor, condition, and market. © The Nova Group / AutoReact.