Sizing a Swiss Target Market from Public Data
You can get a defensible upper bound for a Swiss target market from public registers. ICPs that depend on employee count or industry need an estimate because those fields are unavailable at company level. This guide starts with a concrete Zurich manufacturing ICP and shows where the public count stops. It then explains how to estimate the remaining share.
For a single-account assessment, see sizing a Swiss company from public records.
Worked example: manufacturers in the canton of Zurich
Suppose you sell maintenance and quality software to manufacturers in the canton of Zurich with 20–200 employees. The table applies each filter cumulatively to register data queried on 11 August 2026.
| Step | Companies | Bias introduced |
|---|---|---|
| The register | 791,675 | Nothing yet. Includes dormant shells, holdings and branches. |
| AG and GmbH only | 541,310 | Drops 179,771 sole proprietorships, some of which employ staff and buy software. |
| Registered in the canton of Zurich | 90,311 | This excludes companies registered elsewhere that operate a site in Zurich. |
| Purpose contains "Herstellung" | 7,076 | Creates false positives and false negatives. A holding may list manufacturing activities it never performs. A manufacturer whose purpose says "Handel und Vertrieb" is excluded. |
| 20–200 employees | Not knowable | No public per-company source. Employee-based filtering requires an estimate. |
The observable filters reduce the register from 791,675 entities to 7,076 candidates. Employee count remains unresolved. Any more precise estimate for the 20–200 employee band therefore depends on another source or an explicit estimation method. Ask what data and method produced the figure.
Purpose searches need language-specific terms and synonyms. A Zurich purpose clause is written in German. The French equivalent returns zero companies in Zurich, and the German term returns two in Vaud. Test terms by canton because purpose clauses follow the local register language.
Coverage and limits of public data
Every entity entered in a cantonal commercial register is indexed in Zefix, the federal search service covering all 26 registers. It provides the legal name, legal form, status, registered office, purpose, share capital, first-entry date, and people with signing authority. The same data is published as linked data at ld.admin.ch/query, which is how a whole-population count becomes possible. The register held 791,675 entities on 11 August 2026.
| Legal form | Entities | Share |
|---|---|---|
| GmbH / Sàrl | 290,705 | 36.7% |
| AG / SA | 250,605 | 31.7% |
| Sole proprietorship | 179,771 | 22.7% |
| Foundation | 17,829 | 2.3% |
| Swiss branch office | 15,344 | 1.9% |
| Association | 13,491 | 1.7% |
| General partnership | 11,191 | 1.4% |
| Cooperative | 7,955 | 1.0% |
| Everything else | 4,784 | 0.6% |
For market sizing in the Zefix interface, registered office and legal form are the useful filters. Searching purpose text or a capital range requires another way to query the data.
The UID register, run by the Federal Statistical Office, adds commercial-register status and VAT status, including the start and end dates of VAT liability. It also supports filters for address, commune, and legal form. Of 1,281,879 active UID units on 5 December 2025, 766,447 had a commercial-register entry and 515,432 did not (Federal Statistical Office).
VAT status is a useful screening signal because registration is tied in part to turnover. A Swiss company must register for VAT once annual turnover passes CHF 100,000, but exemptions and voluntary registration affect the result.
Many B2B ICPs start with an industry and a headcount band. Switzerland's public company records do not expose either field per company. Headcount exists in aggregate: the Federal Statistical Office's structural business statistics counted 626,033 market-economy enterprises in 2023, of which 561,952 had fewer than ten employees, 52,476 were small, 9,791 medium, and 1,814 large (BFS, data as of 21 August 2025). About nine in ten enterprises in this dataset have fewer than ten employees. That helps set expectations for the pool. Company-level research is still needed to apply the ICP's threshold.
The Federal Statistical Office uses NOGA codes in its statistics. Public UID company records cover identity, address, register status, and VAT details. Activity code and employee count are unavailable.
| Field | Zefix / commercial register | UID register | Available per company? |
|---|---|---|---|
| Legal form and registered office | Yes | Yes | Yes |
| VAT status | No | Yes | Yes |
| Purpose text | Present in register data | No useful filter in UID | Yes, with language caveats |
| Industry / NOGA | Not exposed as a public company field | Not exposed | No |
| Employee count | No | No | No |
Proxies and their limits
The next step is to use observable fields as proxies and state the bias each one introduces.
| Proxy | Source | Stands in for | How it fails |
|---|---|---|---|
| Share capital | Commercial register | Company size | A CHF 20,000 GmbH can employ two people or two hundred. Low statutory capital separates little. Unusually high capital may be informative. |
| Purpose text | Commercial register | Industry | Purpose clauses are broad legal descriptions and may include activities the company never performs. |
| VAT liability | UID register | Turnover above CHF 100,000 | Binary and set very low. It is a weak activity or scale signal, with important exceptions. |
| Branch count | Commercial register | Multi-site operations | It misses additional sites operated by the same legal entity unless they are separately registered as branches. |
| Officer count | Commercial register | Governance complexity | It reflects governance structure as much as operating size. |
Combining proxies can improve the order in which candidates are reviewed. Employee count remains unresolved. Keep the matched signals beside each company so an editor or salesperson can see why it entered the pool.
From a candidate pool to an estimate
The 7,076-company count is an upper bound. To estimate how many companies fit the full ICP, review a random sample from that pool. Decide whether each company fits the ICP and keep the source behind each decision. Mark cases with insufficient evidence as unclear. Confirmed fits give the lower estimate. Include unclear cases in the upper estimate.
A practical workflow:
- Define the candidate pool with fields the register supports.
- Draw a random sample from the pool. Stratify it by canton or legal form if those groups are likely to behave differently.
- Review each sampled company against the ICP. Record whether it fits and the source used for the decision. Label cases with insufficient evidence as "unclear".
- Use confirmed fits as the lower estimate. Use fits plus unclear cases as the upper estimate.
- Publish the sample size, selection method, review date, and most common reasons for exclusion.
Sizing the market against sizing the list
Keep two outputs: an estimated number of companies that fit the ICP, and a verified list of accounts ready for outreach. The estimate supports planning. The verified list supports execution.
Public registers are good sources for the initial candidate pool because they provide names, identifiers, legal facts, and addresses. Current activity and employee count still require verification. Review a random sample to estimate the market, then qualify individual companies before adding them to an outreach list.
How Prospex approaches it
Prospex helps review the candidate pool by combining register facts with information from company websites and other public sources. Profiles describe the company and its sector. They also identify the customer group it appears to serve and link to the source material. For the example above, those fields help remove companies whose legal purpose mentions manufacturing but whose current business falls outside the ICP.
Automated classifications still need review. Coverage is strongest for companies with a usable public website, and Prospex does not supply an employee count when it lacks a reliable source. Keep unresolved companies marked as unclear in the estimate.
Each candidate keeps the source and review date behind its inclusion. Create a free account to review Swiss company profiles.