How Big Is the Swiss B2B Market? STATENT vs the Register
Swiss market-size estimates often begin with a public register total. Depending on the source, that starting point can range from roughly 632,000 enterprises to more than 1.2 million active UID units. Each figure counts something different.
Prospex approaches a narrower question: how many of the 739,919 register entries that are neither in liquidation nor struck off primarily sell to other businesses? 448,376 of them carry a customer-audience label. After re-weighting the labels by priority band, two calculations produce estimates of about 223,000 and 260,000 primarily B2B sellers.
This estimate covers companies that primarily serve business customers. It is useful when customer audience is part of your ICP. The wider universe of organisations that buy business products falls outside its scope.
The headline number and where it comes from
The Swiss commercial register, indexed federally by Zefix, held 791,675 entities on 11 August 2026. That is a valid entity count. Turning it into a market denominator requires filtering out dissolved companies, holding shells, branches of foreign firms, liquidating entities, and sole proprietorships that never traded.
The UID register is broader still: 1,281,879 active units as of 5 December 2025 (Federal Statistical Office), including entities outside the commercial register. The BFS structural business statistics (STATENT) counted 631,801 market-economy enterprises in 2024. These totals answer different questions. Any market estimate should name the source and its date, then spell out the exclusions. The guide to Swiss company data sources covers the scope and contents of each register.
Register totals are easy to cite because they are public and comprehensive at the entity level. Customer audience, company headcount, and product fit require separate data and filtering.
STATENT versus the commercial register
STATENT, the Federal Statistical Office's structural business statistics, is the official count of enterprises, workplaces and jobs in Switzerland. The latest release describes December 2024 and was published on 20 August 2026. Each release comes out 20 months after its reference month, so a STATENT figure always describes the economy of almost two years earlier.
The count comes from the registers of the AHV compensation funds, completed by the BFS business register. A business enters STATENT once it settles AHV contributions on at least CHF 2,300 of annual income, as an employer or as a self-employed person (BFS survey page, in German). An enterprise is a legally independent unit that takes its own decisions. A workplace is an enterprise, or part of one, at a single location. Employees include the self-employed, and the BFS estimates full-time equivalents with a model built mainly on AHV wage data. Size classes count employees: micro 1–9, small 10–49, medium 50–249, large 250 or more.
| STATENT 2024 | Switzerland |
|---|---|
| Institutional units (enterprises and public administrations) | 649,353 |
| of which market-economy enterprises | 631,801 |
| Micro units (1–9 employees), share of all institutional units | 579,431 (89.2%) |
| Workplaces | 728,380 |
| Employees | 5,681,021 |
| Full-time equivalents | 4,404,860 |
The commercial register counts legal entities that have registered, whether or not they pay anyone. Prospex's copy of the register held 739,919 entries on 26 September 2026 that were neither in liquidation nor struck off. Set canton by canton against STATENT, the two counts diverge in both directions.
| Canton | STATENT 2024: institutional units | Register entries, 26 Sept. 2026 | Register entries per STATENT unit |
|---|---|---|---|
| Zurich | 113,101 | 128,820 | 1.14 |
| Bern | 70,948 | 66,850 | 0.94 |
| Vaud | 59,241 | 65,899 | 1.11 |
| Geneva | 41,260 | 53,973 | 1.31 |
| Ticino | 36,475 | 39,267 | 1.08 |
| Zug | 19,922 | 39,701 | 1.99 |
| Switzerland | 649,353 | 739,919 | 1.14 |
Sources: BFS, STATENT 2024, table px-x-0602010000_106 (institutional units by canton, all divisions). Register: entries neither in liquidation nor struck off, by canton of registered office.
Why the two numbers differ
- Sole proprietors below the threshold: a sole proprietorship must register once its turnover reaches CHF 100,000 a year (art. 931 CO). Farmers and the liberal professions are exempt unless they run a commercially managed business. STATENT counts all of them as soon as they pay AHV contributions, and 49,115 of its 2024 units are farms, forestry or fishing businesses. Bern holds 9,868 of those, which is why it is the one canton in the table where STATENT exceeds the register.
- Dormant and shell companies: a company that pays no wages subject to AHV contributions, such as a dormant AG or a holding that employs nobody, stays in the register and never appears in STATENT.
- Domiciliation addresses: companies registered at a fiduciary's address often have no staff in the canton. Zug has almost two register entries per STATENT unit, and Geneva 1.31. The guide to domiciliation addresses lists the addresses concerned.
- Branches: the register holds 15,044 entries for Swiss branches, each registered in the canton where it operates. STATENT counts a branch as a workplace of its enterprise, which is why it finds 728,380 workplaces for 649,353 units.
- Public bodies and associations: STATENT includes 4,428 public administrations, and an association generally registers only when it runs a commercial business.
- Timing: STATENT describes December 2024, while the register column is a count from September 2026.
STATENT gives counts by canton, sector (NOGA) and size class, with no company names. The register gives names, addresses and signatories, with no sector or headcount. A market estimate usually needs both: STATENT to size a segment, the register to list it. The register carries no NOGA code, so a sector comparison needs a classification on the register side. The target market guide works one through for fiduciaries (NOGA 69.20).
Every STATENT table can be exported from STAT-TAB, the BFS's interactive database, as CSV or Excel. The tables used on this page are px-x-0602010000_106 (canton and NOGA division), _107 (size class), _108 (market and non-market units) and _109 (workplaces). opendata.swiss, the federal open-data portal, also lists STATENT-based datasets from the BFS, cantons and cities.
Register profiles and the historical archive
Prospex stores current register profiles alongside historical SHAB-only records, so its record count is larger than the current register. Of 1,037,412 company records, 800,769 carry a commercial-register profile, within about 1% of the August Zefix total above. The register moves daily, and our copy syncs on its own schedule.
The remaining 236,643 records have no register profile at all. Almost all of them (234,213) are companies we know only from a strike-off notice in the SHAB, the official gazette. These records were struck off before our register sync reached them, so we hold the name, the canton and the deletion, and nothing more.
Prospex takes each profile's legal status from the register. For 3,162 of the 800,769 it is still unknown:
| Status | Records with a register profile | Share |
|---|---|---|
| Active | 736,757 | 92.0% |
| In liquidation | 51,173 | 6.4% |
| Deleted (struck off) | 9,677 | 1.2% |
| Unknown | 3,162 | 0.4% |
The 1,037,412 total describes the Prospex archive. The current register is the Zefix figure of 791,675 above.
The 739,919-entry working denominator
Dropping the entries in liquidation and the struck-off ones leaves 739,919: 736,757 active and 3,162 whose status is still unknown. This is the register count in the STATENT section above, and the set the Prospex market builder starts from unless you change its status filter.
The 739,919 depends on what the register calls active. A company that stopped trading can stay in the register for years before it is struck off. Dormant companies account for part of the 90,566-entry gap with STATENT described above. The true number of operating businesses is smaller.
Customer audience requires separate classification
The register records legal facts such as the company purpose and registered office, together with its signatories. Determining whether a company serves businesses, consumers, or both requires additional evidence. Prospex assigns an audience label from the registered purpose, the company's homepage and its LinkedIn page.
What we actually classified
448,376 of the 739,919 entries (60.6%) carry a customer-audience label: who the company sells to. Each carries one of four labels.
- B2B: the company sells primarily to other businesses.
- Mixed: the company sells to both businesses and consumers (an architect who serves developers and homeowners, a software firm with enterprise and consumer products).
- B2C: the company sells primarily to individual consumers.
- No market: the entity sits outside the commercial seller market (a public authority, a foundation, a membership association, a holding company for asset management). It may still purchase business products, but it sits outside what a B2B sales team would call a market.
The current classifier labelled 433,657 of them in September 2026. It answers two questions separately, whether households are a meaningful customer group and whether businesses or institutions are, and combines the answers into one of the four labels. When its sources leave the answer open, it returns no label. An earlier classifier that read the registered purpose alone supplied 11,162 labels. For 1,877 companies in sectors with a consistent audience, the label comes from the sector.
A separate model also labelled a reference set of companies blind, without seeing the classifier's output. It holds 2,333 labels on operating entries. For 1,680 of those companies, the reference label is the one Prospex publishes. Where the classifier and the blind review both labelled a company, they agree on 414 of 590 (70%). Almost half the disagreements (87 of 176) are companies the review called B2B or B2C and the classifier called mixed.
Audience boundaries are judgment calls. A cleaning company that serves offices and apartments is mixed, while one that serves only offices is B2B. A company that leaves its audience unstated gets classified from its web presence, and the classifier will be wrong for some fraction of those.
Coverage follows the priority score
Prospex gives every company a priority score, a deterministic count of the things that suggest commercial substance: company form, known headcount, web presence, and recent commercial activity. The classifier ran over every operating entry scoring 3 or more, which is also the pool the market builder searches. Below 3, labels come almost only from the earlier classifier and from sectors.
| Priority band | Operating entries | Labelled | Coverage | B2B share of the labelled |
|---|---|---|---|---|
| 10 and above | 26,924 | 24,997 | 92.8% | 49.2% |
| 6 to 9 | 110,184 | 100,898 | 91.6% | 39.3% |
| 3 to 5 | 368,532 | 309,362 | 83.9% | 27.4% |
| 2 and below | 234,279 | 13,119 | 5.6% | 28.1% |
In the three upper bands, almost every unlabelled entry (69,792 of 70,383) is a company where the classifier found the evidence inconclusive. The bottom band is a different population: 175,344 of its 234,279 entries are sole proprietorships, which the score marks down, and 203,952 have no website on record. It holds 32% of the operating entries and 3% of the labels.
The pooled figure
Without adjustment, B2B companies make up 31.3% of the labelled entries.
| Audience | Labelled entries | Share of the labelled |
|---|---|---|
| B2B | 140,298 | 31.3% |
| Mixed (B2B + B2C) | 176,739 | 39.4% |
| B2C | 93,023 | 20.7% |
| No market | 38,316 | 8.5% |
Multiplying 31.3% by 739,919 gives about 232,000, close to the re-weighted 223,000 below, because the bottom band's B2B share (28.1%) is near the pooled one. Re-weighting mainly raises the B2C share: the bottom band holds 32% of the entries but 3% of the labels, and 38.9% of its labels are B2C.
Re-weighting to the operating population
We weight each band's observed shares by the number of operating entries the band actually holds. The bottom band's 28.1% B2B share, for instance, applies to all 234,279 entries in that band. The labelled subset contains 13,119.
| Audience | Pooled share of the labelled | Re-weighted population share | Estimated operating entries |
|---|---|---|---|
| B2B | 31.3% | 30.2% | ~223,000 |
| Mixed (B2B + B2C) | 39.4% | 36.5% | ~270,000 |
| B2C | 20.7% | 26.2% | ~194,000 |
| No market | 8.5% | 7.2% | ~53,000 |
Re-weighting all 448,376 labels gives about 223,000 primarily B2B sellers. Repeating the calculation with the 2,333 blind reference labels alone gives about 260,000. Treat 223,000 to 260,000 as a sensitivity range, or 30% to 35% of the operating entries. It is not a confidence interval. The two calculations split the mixed and pure audiences differently, since the classifier calls a company mixed more readily than the blind review does. B2B and mixed together come to 493,000 to 494,000 in both.
Both calculations assume that labelled and unlabelled companies look alike within each priority band. In the upper bands, the unlabelled entries are the ones whose sources did not settle the question, so they may differ from the labelled ones. The bottom band carries most of the uncertainty, because 13,119 labels stand in for 234,279 entries.
Earlier versions of this guide reported about 104,000 and then 83,000 to 96,000. Both were computed in August 2026, when only 228,845 records carried an active status and about 10,000 had an audience label.
What the numbers mean for a B2B vendor
If your ICP requires prospects that primarily serve other businesses, the re-weighted estimate gives a starting population of roughly 223,000 to 260,000 entries. Including companies with a mixed business and consumer audience expands that pool to about 493,000 to 494,000.
In the market builder, an audience filter removes the companies labelled with another audience and keeps the unlabelled ones. Among the 505,640 entries scoring 3 or more, 136,610 carry a B2B label, 172,882 a mixed one, and 70,383 none.
These figures size the audience filter only. A product TAM also needs sector, headcount, geography, and a reason the company would buy the product. Consumer-facing companies may belong in that market too.
Then add the headcount filter
STATENT 2024 counts 69,922 units with ten or more employees and 13,333 with fifty or more (table px-x-0602010000_107). Set against the 739,919 register entries, that is 9.4% and 1.8%. If the B2B sellers had the same shares, about 21,000 to 25,000 of them would have at least ten employees. This is a rough scenario.
| Starting population | After audience filter | If 9.4% have 10+ employees | If 1.8% have 50+ employees |
|---|---|---|---|
| 739,919 operating | ~223,000–260,000 B2B | ~21,000–25,000 | ~4,000–4,700 |
| 739,919 operating | ~493,000–494,000 B2B + mixed | ~47,000 | ~8,900 |
The headcount distribution within B2B companies probably differs from the register average. B2B share rises with the priority score, and the score rewards known headcount, so B2B sellers may be larger than average. Swiss public sources publish headcount distributions, but no headcount per company. Our guide to sizing a Swiss target market applies the same arithmetic to one segment within the country.
How Prospex approaches this
Prospex lets you filter Swiss companies by the audience they serve: B2B, mixed, B2C, or no commercial market. This is useful when customer audience is part of your ICP. Combine it with sector, company size, and other fit criteria to build the market.
Labels cover 86% of the entries scoring 3 or more, and 5.6% of those below. The automated classification contains errors for some companies. Every profile links to the sources it was built from, so you can check any row yourself.
Build an example market using customer audience and your other ICP filters.