The Companies House data trap: stale records cost UK B2B sales teams real pipeline

Companies House data underpins most UK B2B prospecting tools (Cognism, Apollo, ZoomInfo, internal data warehouses). The data is authoritative for incorporation, registered office, accounts filing dates, director appointments, and PSC declarations. It is NOT authoritative for headcount, revenue, growth rate, or active trading status; those are buyer-self-disclosed via filing extensions or never updated in the public record.

The trap: prospecting tools surface fields like 'employee count' and 'annual revenue' against Companies House records as though they were authoritative. They are typically derived from third-party enrichment (LinkedIn employee counts, financial-data scraping) layered on top of Companies House identity data, with widely varying accuracy.

Three specific traps:

Stale headcount data. A UK SaaS company that grew from 30 to 150 employees in 18 months may show 30 in prospecting tools that haven't refreshed. Outbound sequences targeting them on the assumption they're sub-50 staff misqualify against ICP.

Companies House 'small company' filing exemptions. UK private companies below specific thresholds (turnover under £10.2m, fewer than 50 employees, balance sheet total under £5.1m) can file abridged accounts. Prospecting tools deriving revenue from filed accounts typically show £0 or no data for these companies; the company may be at any revenue point under those thresholds.

Dormant / strike-off-pending status. Companies in dormant status (no active trading) or pending strike-off don't update accounts; the public record looks alive but the company isn't operating. Outbound to dormant or pending-strike-off companies wastes effort.

The implication for UK sales operations teams in 2026: any prospecting tool's revenue / headcount / growth fields should be treated as soft hints, not hard qualification. The hard qualification fields from Companies House (incorporation status, registered office, director changes, PSC) are reliable; the derived fields are noisy. Teams running ICP qualification against derived fields without the soft-hint caveat misqualify at meaningful rates.