Coverage and Quality

Data Bank reports what is supported for the selected release, area, segment and metric. Coverage should be read as part of the result, not treated as an optional footnote.

Geography

Property products cover supported local authorities in England and Wales. They do not currently provide Scotland/Northern Ireland data, postcode-level series, neighbourhood boundaries or arbitrary user-drawn areas. The checked grid contains 318 LADs, but neither that count nor membership should be hard-coded as permanent.

Use /v1/markets/areas for the selected release's codes and names. Geography is release-specific. When comparing differently defined areas across releases, do not assume that an unchanged display name means an unchanged boundary.

Rates are sterling references without an area selector. Do not manufacture a regional rate difference by assigning a national benchmark to an area-labelled request.

Segment and Metric Availability

A supported LAD can have sales data but no reliable new-build differential for a particular type. A rental segment can have rent but no published capital estimate or yield. Estimated price per square metre can be unavailable even when trading counts are present.

Admission is designed to maintain a stable supported universe rather than hide inconvenient months opportunistically. However, a fixed coverage policy is not an unconditional promise of future data. A genuine failed or unsupported estimate must remain unavailable, and future approved coverage changes may occur.

An empty range and a wholly unsupported series are different. Use /coverage and the response's availability, empty_reason, observation status and reason. Never interpolate gaps, carry values forward or replace missing values with zero without labelling that as your own separate calculation.

Individual Months Under Review

An estimate can be withheld for review without withdrawing the whole series or unrelated products. The affected observation retains its period and has status: withheld_pending_review, value: null, a reason and a review_reference. Display it as a gap. The series can still have availability.status: available because other periods contain usable values.

If a capital estimate is withheld, any yield that depends on it is withheld too. The corresponding rent can remain available. A later release may resolve the review; retain release IDs when comparing values. Review references identify the affected observations but do not imply a guaranteed resolution time.

Evidence and Estimation

Recorded activity products summarise an eligible residential transaction population rather than every possible transfer. Transactions can be added, corrected or removed in later source releases. Reporting lag reduces some incompleteness; it does not make records final.

Price-per-square-metre estimates rely on usable property-area evidence linked to transactions. They are not complete observed medians of all sales. Rental and capital estimates describe area/type/bedroom segments, not the condition, specification or achievable price of an individual property.

Delivery estimates infer new-dwelling activity from administrative evidence. Timing, identity and classification coverage can differ from physical completion or sale timing. Development categories describe observed clusters, not verified eventual scheme capacity. Unclassified evidence is disclosed rather than silently assigned to the smallest category.

The intended market scope is ordinary residential property. Administrative classifications and evidence exclusions reduce unsuitable records but cannot guarantee that every specialist or misclassified transaction has been identified.

Rates and Scenarios

Official reference observations, fitted market curves, indicative swap equivalents and hypothetical scenarios have different meanings. A successful API response does not make a fitted curve an executable quote or convert a scenario into a forecast probability.

Historical mortgage spreads can be negative. The spread compares different products and dated references; it is not a pure lender margin. Curve histories can be shorter than scalar-rate histories, and retained scenario snapshots can be much more limited than either.

Precision Is Not Accuracy

Extra decimal places preserve the published value; they do not imply comparable real-world certainty. An estimate displayed to the nearest pound is not thereby accurate to the pound.

The API does not provide a universal confidence percentage for every Data Bank value. New-build differential bounds are modelled uncertainty bounds, not a blanket coverage guarantee for other products. A series' availability.status is not a confidence score.

Refresh and Release Safeguards

Data publication depends on input availability and validation. Failed candidates should leave the last accepted release available with its original dates. Customers should track observation recency and release identity separately from HTTP success.

Service levels, publication commitments and retention terms are governed by your service agreement. Historical coverage and observation dates remain specific to the selected product and release.

Permitted Use and Attribution

Use the data within your account's licence and access permissions. Access to an API response is not, by itself, permission to republish the underlying dataset, redistribute third-party material or represent estimates as official statistics.

Product-specific attribution and redistribution terms will be supplied with the licensed service. These docs do not replace that agreement. Retain appropriate source/release labels and material estimate/scenario limitations in customer-facing reports.