Healthcare services is harder to map from public data than most sectors because the companies in it are regulated, procured, and classified through multiple independent systems that do not share a common entity identifier. A company providing domiciliary care in the East Midlands and a company running independent diagnostics clinics in London will both appear under SIC code 86900 in the Companies House registry. They have almost nothing else in common.
Getting a usable sector map requires integrating three distinct data streams: the Care Quality Commission registration database, NHS and local authority contract award data, and the employment filing patterns visible in Companies House. Each stream covers a different sub-segment with different coverage depth and different lag characteristics. This note describes how we approach the integration and where the method runs into genuine limits.
Why SIC codes fail in healthcare services
The UK SIC 2007 classification groups most private healthcare activity into a small number of codes that mix entirely different business types. Residential care homes, domiciliary care agencies, specialist mental health services, and occupational health providers all sit in adjacent SIC codes with no sub-classification that reflects the operational and competitive differences between them.
These are not similar businesses for deal analysis purposes. A residential care home is a real-estate-anchored, regulated, staffing-intensive business with revenue driven by publicly funded placement rates and occupancy. A domiciliary care agency is a labour-only model with thin margins, high turnover, and revenue tied to local authority contracts on short-term frameworks. A diagnostics business is an appointment-driven operation with high fixed costs in equipment and clinical space. Treating them as a single segment produces a target list that is incoherent to anyone who knows the sector.
The secondary classification layer we need to build for healthcare services therefore has to draw on sources that do not confuse these types, and that requires going beyond Companies House data entirely for the initial segmentation step.
CQC registration data as a sub-segment classifier
The Care Quality Commission maintains a publicly accessible register of all providers registered to deliver regulated activities in England. The register specifies the regulated activity categories that each provider is registered for: personal care, accommodation for persons who require nursing or personal care, treatment of disease or disorder, diagnostic and screening procedures, and others. These categories are operationally meaningful in a way that SIC codes are not.
A provider registered only for personal care is almost certainly a domiciliary care operation. A provider registered for accommodation plus personal care is almost certainly a residential or nursing home operator. A provider registered for treatment of disease or disorder without an accommodation category is likely a clinical services company of some kind. The combination of registration categories, together with the service type and specialism flags in the CQC record, gives us a preliminary segmentation that maps to real competitive groups rather than administrative ones.
CQC data has important scope limitations. It covers regulated activities in England only, not Scotland, Wales, or Northern Ireland, which have separate regulatory frameworks. It does not cover all private healthcare activity: occupational health, cosmetic procedures that do not meet the threshold for CQC registration, and many health technology businesses fall outside the register. And the link between a CQC provider registration and a Companies House legal entity is not always straightforward. A group operator might hold a single CQC provider registration across dozens of registered locations, or might have separate registrations for each trading entity within the group structure.
Resolving CQC providers to Companies House entities
The resolution step from CQC provider to Companies House legal entity requires matching on name, address, and registered location data. For single-site independent operators, the match is usually straightforward: the trading name and registered address align closely with the Companies House record. For group operators, the resolution requires working through the group structure to identify which legal entity holds the operating contract, which holds the real estate, and which is the financial reporting entity that matters for deal analysis.
NHS and local authority contracts are awarded to specific legal entities, and those entities are typically the trading subsidiaries, not the holding company. CQC registrations are held by the provider organisation, which may also be a trading subsidiary. The financial entity we ultimately care about for revenue estimation may be several steps up the corporate structure. We track this by cross-referencing PSC entries and mortgage charges from Companies House to identify the group architecture before attributing revenue signals to the correct entity.
NHS contract data: the revenue signal for publicly funded activity
A substantial portion of private healthcare services revenue derives from contracts with NHS commissioners or local authorities. For companies with meaningful NHS exposure, contract award notices filed on the government's procurement portals provide a direct revenue signal that is not available through any other public source.
The data yield from contract notices varies considerably by sub-segment. Integrated care services contracts, which tend to be large, long-term, and publicly tendered, generate well-documented notice trails with contract values and award dates. Domiciliary care and residential placement contracts, which are often awarded under framework agreements at local authority level, generate less consistent documentation. A company providing significant local authority domiciliary care services may have contract relationships that do not produce a single traceable notice at the contract value level, because the arrangement is a spot-purchase off a framework rather than a discrete awarded contract.
For the sub-segments where contract notice data is available, we use it as a revenue floor rather than a precise estimate. A company that appears across multiple award notices totalling 8 million GBP in a rolling four-year window is not generating 8 million GBP per year from NHS contracts, but it is operating at a scale where NHS revenue is meaningful, and the award history tells you something about its service range and geographic footprint that the balance sheet alone does not.
Employment filing patterns and their relevance to healthcare sub-segments
Employment data from Companies House and job advertisement records is the third layer, and it is where the sub-segment differences are most visible in the filing record. The labour composition of a care home operator, a diagnostics provider, and a mental health services company are structurally different in ways that show up in both job advertisement category distributions and staff cost disclosures.
A care home with a mixed residential and nursing registration will advertise primarily for care assistants, senior care workers, and registered nurses. The ratio of nursing to care assistant posts is an indicator of whether the home operates at the lower complexity personal care end or the higher complexity nursing and dementia end of the residential market. Those two positions attract very different regulatory ratings, staffing cost structures, and fee levels per resident.
A specialist mental health provider will advertise for clinical psychologists, occupational therapists, and mental health nurses alongside managerial staff. The clinical staff composition tells you whether the company is operating low-acuity community-based services or high-acuity inpatient provision. The cost-per-bed and revenue-per-employee profiles for these two types are substantially different.
Labour composition does not give you precise revenue figures. What it gives you, in combination with bed count or location data where available, is a basis for the revenue band estimate that is informed by sub-segment-specific benchmarks rather than a generic healthcare sector average. Applying a domiciliary care revenue-per-employee benchmark to a diagnostics company would produce a wildly inaccurate estimate. The preliminary segmentation from CQC data is what makes it possible to apply the right benchmark for each company.
Integration: where the sources overlap and where they do not
The three data streams cover the sector with different emphasis. CQC data is comprehensive for regulated activity providers in England but says nothing about companies outside that scope. NHS contract data covers publicly funded providers but is thin or absent for companies operating primarily in the private-pay market. Employment filing data is available for all incorporated entities but requires sub-segment-specific interpretation to be meaningful.
The integration logic we use starts with the CQC register as the basis for the sector universe, resolved to Companies House entities. We then append contract data where available as a revenue and customer type signal. Employment data is then layered on as a size estimation input, calibrated against the sub-segment classification established in the first step. Companies that have no CQC registration but appear under healthcare-adjacent SIC codes are handled separately, using employment and contract signals alone, with the segmentation driven by the job category composition of their hiring history.
The result is a sector map that is more granular than anything a SIC-code filter can produce, but it is not complete. Sub-sectors with low public funding exposure and limited regulatory registration requirements, including many health technology and health analytics businesses, are partially visible in the map but not fully covered by any of these sources. For those sub-segments, the filing record provides structural information but not reliable revenue signals. We identify those gaps explicitly in the output rather than extrapolating coverage we do not have.
Where the limits genuinely sit
Three limitations are worth naming directly. First, the CQC-to-Companies-House resolution fails for some group operators where the legal structure is complex or where the operating entity is registered under a name that differs significantly from the trading name. The error rate is low but not zero, and affected entries are flagged for manual review rather than included with standard confidence scores.
Second, the NHS contract data has a geographic and commissioner-level coverage gap. Framework agreements at ICS or local authority level are not uniformly published in a way that generates traceable contract notices at the provider level. Companies with predominantly local authority revenue in certain regions will appear undersized in the contract layer relative to their actual activity.
Third, the employment signals lag significantly for companies that last recruited in 2021 or 2022 during the post-pandemic workforce expansion. Their current workforce may be substantially different from what the job advertisement record suggests, and their Companies House filing headcount may reflect a period of unusually high or low employment that distorts the size estimate. For companies where the employment signal is old and the accounts are more current, we weight the account-based balance sheet signals more heavily than the employment proxy.
None of these limitations make the mapping approach unreliable for its intended purpose, which is generating a prioritised target list for a deal team assessing a specific sub-segment of UK healthcare services. They do make it important to match the method to the use case: a fund looking at residential care operators in the 5 to 20 million GBP revenue range will get reliable output from this approach; a fund looking at early-stage health technology businesses will find the coverage thinner and should not treat the map as exhaustive.