Comparing mold risk between two properties
How to compare mold risk between two properties: score both, then compare the drivers rather than the numbers, and check data completeness before treating a gap as real.
Comparing by score vs Comparing by driver
| Axis | Comparing by score | Comparing by driver |
|---|---|---|
| What it actually measures | One number per property, which sorts them. | What produced each number: which module carries the weight, and whether it is documented or inferred. |
| What it cannot see | Whether the two scores rest on comparable information. A 55 from verified records and a 55 from a neighbourhood median are not the same claim. | Nothing much — this is the comparison that survives scrutiny. |
| Cost range (USD) | $0 for both. | $0 for both. |
| Time to result | Seconds each. | Seconds each, plus the few minutes it takes to read them side by side. |
| When it is required or accepted | Never. | Never. |
| Who accepts it (lender / insurer / court / buyer) | By nobody as evidence. | By nobody as evidence — but it is the version that informs a decision. |
A four-point gap is noise; a different driver is a different building
Two properties scoring 58 and 54 are, for practical purposes, the same. Two properties both scoring 58 — one carried by regional climate, the other by an unresolved documented water event — are not remotely the same, and the number hides that completely. Compare what produced the score, and check completeness before treating any gap as meaningful.
The order to compare in
Data completeness first. If one property has a verified construction year, a parcel record and a filed disclosure and the other has a postal code and a census-tract median, you are not comparing two estimates — you are comparing an estimate and a guess. Close that gap before anything else, usually by getting the construction year confirmed.
Then the module carrying the weight. Regional load applies to everything in a postal code and is not a difference between two properties in the same area. Terrain, building and documented history are where genuine differences live.
Then evidence strength. A driver marked confirmed is a fact about the building; one marked inferred is a statement about its cohort; one marked suspected is close to a guess. Two properties with the same driver at different strengths are in different positions.
Then what is fixable. A property carried by site drainage and ventilation has a cheap path to a better building. One carried by construction era, cladding and terrain does not.
Where comparisons go wrong
Comparing across model versions. Every score carries the version that produced it, and a difference between an old and a new score may be the buildings or may be the weights. Re-run both on the current version before reading anything into a gap.
Comparing a property with a parsed disclosure against one without. The disclosure module is 45 of 100 points, and a property where it was not analysed is not scoring lower on it — it is not scoring on it at all. The report says which modules were analysed, and that line matters most in a comparison.
Treating a low score as the safer purchase. It may simply be the property nobody has recorded anything about.
Comparing more than two
Across a shortlist the free check does the ordering, and that is the right use — it costs nothing and it needs no access to any of them.
Buy the paid report on the one or two you are actually pursuing, and only where a disclosure exists to parse.
Then send an inspector to one, with the driver list. Sending inspectors to four is not a sensible use of anyone's money, and choosing which one to send them to is what this whole exercise is for.
The basis for these figures
General US ranges for planning, not quotes.
Questions people ask
- How big a score difference actually matters?
- A band change usually does; a few points usually does not. The bands exist precisely because precision at the single-point level would be dishonest about what records can support.
- Can I compare properties in different states?
- Yes, with a caveat: full four-module scoring runs in Pennsylvania, New Jersey and Delaware. Elsewhere the disclosure module is unavailable, so the two scores rest on different amounts of the model.
- Does it cost anything to compare?
- No. The free check runs on any address with no account, so a shortlist costs nothing to order.
- One property has a disclosed water event. Is it automatically worse?
- Not necessarily. A disclosed and properly remediated event, with a scope and an independent clearance, is a known quantity. A property with no history at all is an unknown one, and only one of those two can be priced.