# How to research a property’s estimated value

_By Apoorv Khanna, September 28, 2026_

Research a property's estimated value by checking the subject record, inspecting the comparables and keeping each price's date and meaning attached. An automated estimate can help organize the investigation. It cannot observe every repair, title issue or change in condition that affects a real transaction.

The worksheet below is for research. It keeps asking prices, documented sale prices, tax assessments and model estimates in separate fields so an agent cannot accidentally treat them as interchangeable.

## Establish the property you mean

Supply the complete address, including unit, and the relevant property type. State whether you are researching a whole building or one unit. Add any reliable information about floor area, bedrooms, bathrooms and condition, with its source.

After [connecting Vaaya](/install), use:

```text
Research the estimated value of [complete property address].
Purpose: [research question]. Known facts and sources: [details].
Confirm the subject record before requesting an estimate.
Show the estimate range and comparable records, including price type and date.
Keep missing condition or transaction details explicit.
Use [total research budget], with a ceiling on each call.
Return a worksheet and questions for a local professional, not a purchase offer.
```

The current services described here focus on US property data. Check coverage for the specific address before treating an empty result as evidence that the property does not exist.

## Inspect the inputs and comparable records

1. **Retrieve the subject record.** Ask `consult` for a suitable property lookup, such as `rentcast/properties`. Match address, unit and property type. Compare available floor area and room counts with the user's supplied sources; preserve discrepancies rather than selecting whichever value gives the preferred estimate.

2. **Request one appropriate estimate.** `rentcast/value-estimate` accepts an address or coordinates, with optional subject attributes and comparable-selection parameters. Confirm the current schema and quote before running. Save the exact inputs, including any overrides, beside the result.

3. **Read the estimate and its basis.** RentCast's [valuation documentation](https://developers.rentcast.io/reference/property-valuation) says its model uses comparable listing information and returns the subject, range and comparables. Keep the range visible. Its [response schema](https://developers.rentcast.io/reference/property-valuation-schema) identifies the relevant fields; do not assume a low and high value represent guaranteed sale bounds.

4. **Review the comparables individually.** Check location, property type, size, relevant date and listing status. Mark obvious mismatches and explain them. If the question requires completed sale records, request a source that identifies sales explicitly, such as an appropriate RealEstateAPI comparable-sales lookup. Verify the date and transaction fields in each returned record.

5. **Write the research conclusion with its unresolved inputs.** Describe where the subject sits relative to the observed records and which missing facts could change that assessment. Explain differences between sources before considering another paid estimate. Two outputs may share underlying data and are not necessarily independent opinions.

## Keep the worksheet separated by evidence type

Use these fields for the subject and estimate:

```text
subject_address | unit | property_type | source | retrieved_at
area | area_unit | bedrooms | bathrooms | condition_source
estimate | estimate_low | estimate_high | currency | model_provider
input_overrides | comparable_selection | unresolved_subject_facts
```

For comparables, record `source_url_or_record_id`, `price_type`, `price`, `transaction_or_listing_date`, `distance`, `size`, `condition_evidence` and `reason_included`. A stale listing with an unknown closing outcome can remain context, but should not be relabeled a recent sale.

Avoid averaging all the visible prices into a new authoritative-looking number. Combining a tax assessment, an asking price and a model estimate hides their different purposes. If you calculate a descriptive median for a comparable group, retain the inclusion rules and record count.

## Change assumptions deliberately

An estimate can fail because too few eligible comparables exist. Widening a radius or lookback period may return more records, but also changes the question. Save each version and explain which constraint was relaxed.

Track subject lookup, estimate and optional corroboration against one research allowance. Reuse records that already answer the question. End with the specific evidence needed next: verified floor area, repair scope, a confirmed transaction record or a local professional's assessment. Bring the source-linked worksheet to that review rather than treating the model's point estimate as a final price.

## Questions

**Is an automated value estimate a professional appraisal?**

No. Treat it as a model estimate based on the provider’s data and inputs. A consequential transaction may need a qualified local professional and additional evidence.

**Are all comparables completed sales?**

No. Some services use comparable listings. Preserve the record type, price basis and date so asking prices are not presented as closed sale prices.

**What if the property details are wrong?**

Resolve the subject identity and document corrected attributes before rerunning. Keep the original inputs and explain the change rather than silently replacing them.