UK BUSINESS STRUCTURE · 13 March 2026 snapshot
A small share of registered UK businesses operates a much larger share of business locations. Calculations from the ONS's 2026 register snapshot show that multi-site enterprises represent about 2.1% of businesses but account for 15.6% of local units. Counting businesses and counting sites answers different economic questions.
The release, published on 24 September, records 2,786,985 enterprises and 3,232,415 local units. A local unit means an individual site belonging to an enterprise, such as a shop or factory. The figures cover businesses registered for Value Added Tax (VAT) and/or Pay As You Earn (PAYE), rather than every form of self-employed activity in the UK. ONS provides the source counts in Table 5.
Which denominator makes the difference?
Multi-site share of all enterprises
Multi-site share of all local units
| Sites per enterprise | Enterprises | Local units |
|---|---|---|
| 1 | 2,729,010 | 2,729,010 |
| 2–4 | 45,290 | 108,015 |
| 5–9 | 6,970 | 44,365 |
| 10–19 | 2,960 | 39,335 |
| 20 or more | 2,755 | 311,690 |
| Total | 2,786,985 | 3,232,415 |
The single-site row contributes exactly one local unit per enterprise. The remaining rows create the gap between the totals. A ratio of all local units to all enterprises gives about 1.16 sites per business. That overall average hides the difference between a very large single-site group and a small group operating networks of locations.
Reproduce the multi-site shares and average
Add the four multi-site enterprise rows: 45,290 + 6,970 + 2,960 + 2,755 = 57,975. Their local units total 108,015 + 44,365 + 39,335 + 311,690 = 503,405.
The enterprise share is 57,975 ÷ 2,786,985 × 100 = 2.080%. The local-unit share is 503,405 ÷ 3,232,415 × 100 = 15.574%. Within that group, 503,405 ÷ 57,975 gives approximately 8.68 sites per enterprise. Published counts are rounded, so these derived values should not be treated as more precise than their inputs.
Why does this matter for local economic analysis?
Consider a hypothetical business with one headquarters and nine other locations. An enterprise count records one organizational unit; a local-unit count records ten sites. Opening an eleventh site could expand its physical footprint without adding an enterprise. A change in ownership could also alter the organizational count without an equivalent change in the places where activity happens.
Those examples show why a business-count chart cannot, by itself, establish how many workplaces opened or closed. They also explain why local comparisons need care. ONS generally allocates a multi-site enterprise to its main operating site or head office, while local units describe individual locations. The quality and methodology report explains the geographic treatment.
Does a high site share mean a high market share?
No. A small office and a large factory each count as one local unit. Site counts contain no common measure of floor area, capacity, staffing or revenue. An enterprise with many locations can have a different economic weight from another with fewer, larger locations.
For a market-concentration question, first specify the market and the relevant sales or output measure. For a workplace-footprint question, local-unit counts can be useful, provided the geography and date match the question. Keep these two uses separate in a dashboard.
What does the snapshot leave out?
Unregistered low-turnover, non-employing businesses fall outside this register-based publication. Multiple registrations at one address can distort very local comparisons, and the snapshot describes March rather than current October conditions. The release also notes a change in industry coding, so detailed sector movements deserve a method check.
The practical result is a clearer unit of analysis: use enterprises to study organizations and local units to study their locations. Neither count automatically measures entrepreneurs, independent owners, jobs created or economic value.
Sources, dates and reuse
ONS, UK business; activity, size and location: 2026, Table 5, released 24 September 2026. Observation date: 13 March 2026. Definitions and coverage: quality and methodology report, revised 24 September 2026. Source check: 9 October 2026, 01:57 UTC. All comparisons and graphics are original; chart values are repeated in text or tables. Source: Office for National Statistics, licensed under the Open Government Licence v3.0. ONS reuse guidance permits commercial reuse with attribution, subject to exceptions. No agency logo, external photograph or third-party chart is reproduced; no official endorsement is implied.
