CraftrUSA Field Notes
How to Design Property Scoring Software Without Hiding the Evidence
A score is useful only if a user can understand what drove it. Property intelligence software should simplify the decision without erasing the observations and evidence behind that decision.
Published 2026-10-03 • Updated 2026-10-03
Separate the property from the assessment
The property is the long-lived record. An assessment is a point-in-time evaluation. Keeping those concepts separate allows the system to support reassessment, history, changing evidence and future monitoring without overwriting prior findings.
Store evidence before calculating the score
Photos, observations, device readings, user answers and professional findings should retain their source and confidence. The scoring layer can then interpret those records while the underlying evidence remains available for review.
- Observation
- Source
- Date
- Confidence
- Category
- Severity or condition
- Supporting media or data
Recommendations should be traceable
A recommendation is more credible when the application can show which findings caused it and whether the recommendation is educational, preventive or urgent. AI can help organize and explain recommendations, but the application should preserve the relationship to the source findings.
Reports are a product surface, not just an export
A report should translate technical observations into priorities a homeowner or professional can act on. The report model should be designed alongside the scoring model so the system does not end with a technically correct score and an unusable explanation.
What Home Health Scan demonstrates
Home Health Scan applies this architecture to healthy-home assessment. The product model organizes property-based evidence, a 0–100 Home Health Score framework, recommendations and report-oriented outputs, with room for reassessment and monitoring over time.