Methodology
How Synthesis Data builds a property-intelligence brief
The objective is not to turn every data point into a conclusion. The objective is to preserve the source trail, identify what the evidence supports and make limitations visible.
Evidence classes
Source-derived fact
A statement traceable to a retained source, with publisher, retrieval date, data period and applicable geography when relevant.
Analysis
A reasoned interpretation based on cited evidence. Analysis is not presented as a source fact.
Estimate or scenario
A calculated or hypothetical result with assumptions disclosed. It is not an appraisal, quote, Loan Estimate, insurance determination or guarantee.
Unresolved item
A question that cannot be supported adequately by the available evidence. It remains unresolved rather than being filled with synthetic output.
Minimum provenance record
For time-sensitive or property-specific evidence, the working record should retain the source reference, source owner or publisher, retrieval timestamp, observed/data period, geographic scope, applicable terms or authorization status, and a content hash or other durable evidence reference where appropriate.
Human review
Automated collection and AI-assisted drafting may support research workflows, but material client-facing conclusions remain subject to the applicable human QA and approval controls. AI-generated text does not convert an unsupported statement into verified evidence.
High-risk subjects
Mortgage, lending, TRID, RESPA, appraisal, insurance, tax, zoning, licensing, legal, fair-housing and regulated representation questions require source-specific qualification and, where appropriate, review by the responsible licensed or compliance professional.
What this methodology does not do
Synthesis Data does not issue a Loan Estimate, appraisal, insurance coverage determination, underwriting decision, title opinion, legal opinion or guaranteed investment result through this research methodology.