A populated property field is not automatically a confirmed fact. A useful review asks where the value came from, when it was observed, how directly the source supports it, and whether stronger evidence is available.

The objective is not to rank every source as good or bad. It is to understand what each field can support in the current underwriting workflow.

Label the kind of information first

Start by separating fields into five practical categories:

  • User-supplied: entered by the insured, agent, broker, or another participant in the submission.
  • Sourced: returned from a public or licensed record, such as an assessor or permit source.
  • Observed: visible in imagery, an inspection, or another direct observation.
  • Estimated: calculated from stated assumptions or a model, such as replacement-cost context.
  • Unavailable: not returned, not covered, or not reliable enough to present.

These labels describe how a value entered the file. They do not determine whether it is correct. A user-supplied roof year may be supported by invoices, while a sourced field may lag a recent renovation.

Read field-level provenance

A source name attached to the whole report is less useful than source context attached to individual fields. For a material input, try to preserve:

  • the source or source type;
  • the source record date or observation date, when available;
  • the retrieval date;
  • the original value;
  • any normalization or mapping applied; and
  • whether a user later corrected or overrode it.

This is especially important for construction, occupancy, building area, year built, roof characteristics, protection, permits, and valuation inputs. Those fields may come from different records and may not share the same date.

Treat dates as part of the value

Recency matters differently by field. A building’s original year built is usually stable. Occupancy, roof condition, surrounding vegetation, nearby construction, and protection features can change.

Do not treat a recent retrieval date as a recent observation date. A record downloaded today may still describe conditions from years earlier. When the underlying date is unavailable, label that limitation rather than implying freshness.

For imagery, review the capture date and the portion of the property that is visible. For permits, distinguish application, issue, inspection, completion, and record-update dates when the source provides them.

Use confidence as a routing signal

Confidence should help decide what to review next. It should not turn uncertainty into a precise-looking fact.

Higher confidence may reflect direct evidence, agreement among independent sources, a strong property match, or a recent observation. Lower confidence may reflect an incomplete address, an old record, a broad model assumption, or disagreement between sources.

Ask what drives the confidence label or range. A score without its basis can hide the very uncertainty the reviewer needs to see.

Reconcile conflicts without averaging them away

When two sources disagree, first test whether they describe the same thing. Building area may mean gross area, living area, rentable area, or footprint. Construction may describe the structural frame in one system and the exterior wall in another.

Use this sequence:

  1. Confirm the property and building match.
  2. Compare field definitions and units.
  3. Compare observation or record dates.
  4. Check for renovation, expansion, demolition, or a change of use.
  5. Prefer stronger evidence for the specific question, and preserve the conflict.

Do not average categorical fields or use the newest date automatically. A newer source can repeat an older error.

Match the evidence to the decision

Different questions require different evidence. Aerial imagery may help identify a roof shape or nearby exposure, but it may not establish roof-covering material or condition. A permit can indicate that work was authorized, but the record may not prove completion. An assessor record may help identify size and use while still lagging a recent addition.

For visual review, see how to use aerial imagery in property underwriting. For permit evidence, see what permit history can and cannot tell you.

Document overrides

When better information replaces an existing value, retain the prior value and record the reason for the override. Useful notes identify the evidence, reviewer, and date rather than simply marking a field “corrected.”

Overrides should also trigger dependent calculations when appropriate. A change to building area, construction, occupancy, or roof information may affect valuation context or peril factors. The original output and the revised output should remain distinguishable.

A practical field review

For each material field, ask:

  • What exactly is being described?
  • Who supplied or observed it?
  • When was it true?
  • Is the property match reliable?
  • Does another source disagree?
  • What decision will use this field?
  • What evidence would resolve the uncertainty?

A transparent property file does not pretend that every input is equally certain. It gives the next reviewer enough context to use each value appropriately and to recognize when a follow-up question is warranted.

Apply the same field-by-field review to the sample report: check property match, source, date, confidence, and what evidence would resolve an open question.

For an example of why dates belong with source records, the U.S. Geological Survey’s aerial-photo metadata dictionary defines an acquisition date as the day the photography was acquired. Preserve a source’s own date field and meaning instead of substituting a page’s update date.

Your first 10 full reports are free.

Run a U.S. address check to see a preview, then use your work email to unlock the full free report. No credit card required.