Swiss Directorship Concentration: What a Year of Officer Changes Shows
One director appears across 40 companies in your prospect data. The count may point to a fiduciary's book of administrative entities. It may also reflect an operating group. In other cases, several unrelated people share a name. The registered addresses and business activities help you decide what to review.
This analysis covers officer changes published in SHAB during the 12 months ending 27 July 2026. It contains 126,409 publications involving at least one appointment or departure at 99,888 companies. It measures recent changes. The register provides the complete set of board and management mandates.
Coverage of the 12-month change window
A publication appears when something changes. A trustee who already sits on twenty boards and moved none of them during the year is absent from this data entirely. The register holds 540,630 AGs and GmbHs. Of these, 81,150, or 15.0%, published an officer change in the window. The other 85% contributed nothing, and among them are most of the long-standing seats.
These filings are useful for spotting recently added or removed mandates. They miss directors whose mandates stayed unchanged during the window, including professional directors with a stable portfolio.
What one name looks like across the year
The 126,409 publications contain 284,646 named parties. Of these, 18,121 are companies: auditors and corporate shareholders, excluded here. The remaining person records form 189,666 distinct name strings and 253,332 unique name-company pairs.
| Companies per name | Distinct name strings | Share of names |
|---|---|---|
| 1 | 157,100 | 82.8% |
| 2–4 | 29,017 | 15.3% |
| 5–9 | 2,944 | 1.6% |
| 10–19 | 502 | 0.27% |
| 20–49 | 101 | 0.05% |
| 50 or more | 2 | 0.001% |
Within this 12-month change dataset, the most frequent name appears at 84 companies. Two names appear at 50 or more. These figures describe recent filings, so they cannot establish the maximum number of mandates held at one time.
Repeated names still create a meaningful review pool. Of the 94,480 companies in this window that name at least one natural person, 17,586 carry a name that also appears at four or more others: 18.6% of them, reached by just 3,549 names. Narrow it to the 605 names seen at ten companies or more, and you still cover 6,542 companies. First identify merged names. Then separate administrative clusters from operating groups.
Three patterns behind a high count
A count can produce a shortlist, but it cannot classify the companies by itself. These examples show three patterns worth checking.
A domiciliation book
In Lucerne, 344 register entries are written care of a single service company, and 300 of them published an officer change during the year. Inside that one cluster, 21 different names appear at ten or more entries, and the ten most frequent names between them account for 235 of the 344. The concentration points to an administrative-services cluster. Check the entities for staff and current business activity before sending them to contact research. This is the pattern the domiciliation-address guide covers from the address side.
An operating group
Two names appear together on 45 of the same companies. Of the 50 companies involved, 46 are dental practices, spread across 13 cantons and 44 distinct addresses. The repeated sector and broad address spread point to an operating clinic group. These companies should stay in the review pool. The next question is whether purchasing happens at each practice or at group level.
Several people with one name
A common given name and surname can combine several people under one string. One such name reaches 27 companies at 26 different addresses across 10 cantons: a large bank, two Migros subsidiaries, a regional hospital, a housing cooperative, a joinery and a bakers' training fund. The sectors and locations show that this record needs to be split before the count can guide a decision.
Address concentration suggests the next check
Take the 605 names that appear at ten or more companies and look at where those companies are registered. For 209 of them, 35%, half or more of the companies sit at one address. The median name string in the group has 36% of its companies at its single busiest address.
That concentration often points to a fiduciary or administrative cluster. A dispersed pattern calls for a sector check. Repeated business activity may reveal a group, while unrelated activities may indicate a name collision.
For 94 of the 605 name strings, at least 90% of the registered addresses are distinct. Those cases are a useful starting point for checking name collisions.
SHAB has no stable person ID
SHAB publishes a person's name and two location fields: place of origin and place of residence. It does not assign a stable public person ID. This analysis joins records by name alone: the origin and residence fields cannot be matched consistently across filings.
Name matching creates errors in both directions. A common name can merge several people into one count, inflating it while looking exactly like a genuine professional director. Extra given names or spelling changes can split one person's mandates across several counts.
One name in this data appears at 37 companies, and a longer spelling of it, carrying two extra given names, at another 33. The two sets share no company at all, yet 32 of the 70 are registered care of the same corporate-services firm and 25 sit at one address in Zug. That overlap makes it likely that both spellings refer to one person. Confirm the match manually before combining the records.
Use name counts to prioritise review. For high-count records, compare addresses and business activities before merging people or excluding companies.
Coverage is uneven across cantons
Cantons publish officer changes in two different formats. Most of them use labelled blocks, which parse cleanly. Geneva, Vaud and Neuchâtel write the same acts as running prose, and prose is harder to read reliably. A sentence such as "Dupont Marie n'est plus gérante; ses pouvoirs sont radiés" carries the same facts in unlabelled prose.
The gap shows up in the yield. Across the cantons with labelled blocks, 40.4% of SHAB publications produce an officer-change record (105,888 of 262,281). In Geneva, Vaud and Neuchâtel the figure is 32.2% (20,521 of 63,711). Some of that difference is a genuine difference in what those cantons publish, and some of it is extraction that misses acts a labelled format would have handed over. Nothing in the data separates the two.
This data cannot rank cantons by director concentration. A canton that parses less cleanly will look calmer than it is. Use the counts within a canton, where the extraction is consistent, and treat any comparison across extraction formats with caution.
Working it into a list
- Count companies per name string. Use five companies as an initial review threshold and keep the raw count visible.
- Group each flagged record by registered address. A dominant address raises the likelihood of an administrative cluster. Check the c/o line and the address occupant. Then look for evidence of current business activity.
- Review dispersed records by business activity. A repeated sector can reveal an operating group. A broad mix of unrelated activities suggests that several people share the name.
- Choose the sales unit. For an operating group, identify whether the local entities buy independently or whether procurement sits at headquarters. Route uncertain matches to review, where each company in the cluster remains available for an individual decision.
How Prospex handles it
Prospex combines officer concentration with evidence from the company's website and other public sources. A repeated name can raise a review signal. Clear operating activity can keep a company in the result even when its officers also appear elsewhere.
When Prospex finds no website or discoverable activity, that absence weighs against the company. The reason remains visible on the row, so the user can review the decision and restore a company that was classified incorrectly. See an example market built from the same public-register sources.