Inconsistency between Officer Bulk Snapshot and Stream: Incorrect officer_role and Address Formatting

Data Issues
1

We have noticed significant inconsistencies between the Officer Bulk Snapshot and the Streaming API data. The Streaming API (and CH Service) appears to be correct, but the Bulk Snapshot contains conflicting information for the same records.

We have identified approximately 100,000 inconsistent records in recent snapshots.

The Issues:

  1. Officer Role Mismatch: In the snapshot, the officer_role is listed as secretary (or other natural person roles) even when is_corporate is true. In the Stream, the same record correctly shows corporate-secretary.

  2. Address Formatting: address_line_1 often has the building number and street name swapped between the two sources (e.g., “16 HEATHER GARDENS” vs “HEATHER GARDENS 16”).

Example Data (Company SC835337):

Source officer_role is_corporate address_line_1
Snapshot (2026-02-03) secretary true 16 HEATHER GARDENS
Stream (2026-02-01) corporate-secretary true HEATHER GARDENS 16

Raw Data Snippet:

company_number  officer_role        is_corporate  address_line_1       data_source
SC835337        00 (secretary)           Y          16 HEATHER GARDENS   snapshot (2026-02-03)
SC835337        corporate-secretary true          HEATHER GARDENS 16   stream (2026-02-01)

The is_corporate flag seems to be set correctly in the snapshot (true), but the officer_role string contradicts it. The Stream data appears to be the accurate version.

Could you please investigate why the Bulk Snapshot is generating these inconsistent roles and address formats?

Thanks.

2

Not from CH but two observations

  1. In Prod 195 the Officer role is a natural person UNLESS the corporate indicator (Y” if the officer is a corporate body, otherwise BLANK) is previously present in the row.
  2. Address Formatting issues as described are usually the outcome of the (poor) original entry quality. Proper entry of addresses is very clearly beyond the capability of so many nowadays.

StatBooks® - the established Companies House data wranglers