Finding Companies Like Your Best Customers in Switzerland

11 Aug 2026 7 min read

Pick the customers your team would happily win again. They had a clear reason to buy and remain a good fit after the sale. Finding more Swiss companies like them sounds simple. Registered purposes and broad industry labels bring back plenty of false positives, so the seed group and the review matter as much as the matching method.

Start with customers that bought for the same reason. Define the resemblance you care about and build a candidate list from commercial descriptions. Test a sample before sales uses it.

What "similar" has to mean

Two companies are commercially similar when the same problem and offer are relevant to both. Start with the business itself:

What it sells. Use the detail a customer would recognise. "Software" is too broad. "Shift planning for hospitals" gives the search something useful to match.
Who buys from it. A supplier to Swiss municipalities faces a different market from a supplier to global pharmaceutical groups, even when their technology looks similar.
How it operates. Delivery model, sales model, regulated environment and procurement route can change whether your offer fits.
Why your customer bought. Record the problem your offer solved. Customers with different reasons to buy belong in separate seed groups.

Use growth to help order a list that already reflects commercial fit. A new operations lead or compliance deadline can also lift a company's priority.

Group customers by the reason they bought. A hospital group and a watchmaker can both be valuable accounts while sharing little that helps a similarity search. Build a separate seed group for each repeatable sales motion. Three coherent examples will usually produce a more useful brief than a long mixed list.

Build the first list

Write the same four lines for each seed customer: what it sells, who buys, how it operates and why your product fitted. Keep only the details shared across the group. Brand names and one-off projects will make the search brittle. Keep temporary buying signals for later ranking.

Turn the shared details into a two-sentence search brief. For example: "Swiss providers of shift-planning software for hospitals and care groups. The product is bought by operations or workforce-planning teams and must fit regulated staffing processes."

Use that brief to collect candidates from company websites. Specialist directories and association lists can add candidates. Compare offer and buyer first. Apply hard constraints such as geography or legal form afterwards, then use current buying signals to order the qualified companies. Build keyword searches in German, French and Italian, then check the result by language region. One-language terms will skew the list even when the underlying market is national.

Why registered purposes make noisy matches

A registered purpose gives every search a common source, but its aim is to cover the company's legal scope. Prospecting needs a description of the operating business. Broad permissions and repeated legal clauses can outweigh the few words that provide one.

In an internal sample of 4,000 random company pairs, registered purposes produced far more similarity between unrelated businesses than website-based commercial summaries. This is evidence about ranking quality. It does not quantify how alike two companies are.

The purpose clause gives the company room to manoeuvre and borrows standard formulas from the notary's template. The closing boilerplate often covers acquiring holdings and assets, including real estate and intellectual property. Shared legal wording can lift unrelated companies in the ranking. A company that builds bioprocess plants draws pharmaceutical manufacturers and clinical-stage biotechs when matched on purpose text. Matching commercial summaries surfaces firms that build the same equipment.

Where NOGA helps, and where it blurs the market

NOGA is a useful starting filter when the main activity matches the reason your customers buy. It narrows a market by principal output, but a code-based search has limits in Switzerland.

No public per-company code. Public UID entries do not display a NOGA code. Commercial datasets may supply one, so check how the provider assigned it and whether it represents the company's current main activity.
Codes group by output, buyers by need. A precision machining shop and a medical-device assembler sit in different classes and buy the same quality software.
A main code leaves secondary work out. The activity that matters to your offer may sit outside the company's primary classification.

Testing the list before you work it

Whatever method produced your list, it can be tested in an afternoon, and the test is cheaper than a month of calls into a bad one.

Hold back customers you already know. Build the seed from one coherent group and keep a few similar customers out of it. Their position in the results gives you an early check. Review the other results as well, because a handful of known accounts cannot measure the whole list.
Review the top and the cutoff. Check ten companies near the top and ten around the planned cutoff. Add ten just below it. Record the offer match, buyer match, operating context and reason for exclusion. Move the cutoff or revise the seed when the pattern changes.
Check regional balance. Compare the list with the market you expected. A strong cantonal concentration may come from the seed or the source coverage. It may also reflect the market itself. Inspect the cause before changing the ranking.

Review false positives before excluding them

Four categories deserve a closer review before outreach.

Competitors. A competitor of your customer may be a strong prospect because it serves a similar market. A competitor of your own company belongs on the exclusion list. Identify the relationship before removing it.
Suppliers and integrators. Keep them when they match the buyer profile. Route genuine partner candidates to the person who handles partnerships.
Holding and property vehicles. Check the website, employee presence, operating address and role within the group. Remove entities with no relevant buying function.
Public bodies and associations. Keep them when your team serves the public or membership sector. Otherwise exclude the relevant entity types at the start.

Broad exclusions can remove valid buyers along with false positives. A firm that does the work your product supports is usually a buyer. Agencies running paid campaigns buy campaign tooling, and accountancy practices do the same with accounting software.

Use exclusions for relationships and entity types that clearly sit outside your sales motion. Rank ambiguous cases lower and review them individually. Broad category exclusions can remove the buyers most likely to understand your product.

How Prospex does this

Prospex compares companies using short descriptions built from their websites and public sources. Each description focuses on what the company sells and who buys it. Start with a small group of customers that bought for the same reason. Discover ranks companies with similar commercial descriptions, and each company page shows related businesses.

Review the ranking before outreach. Apply your market constraints and remove seller-side competitors. Then use current signals to order the remaining companies. The similarity score reflects the commercial-description match.

Coverage is strongest among companies with an identifiable and informative website. When Prospex has no commercial summary, it falls back to the registered purpose, so those matches need closer review. A company without enough usable source text may be absent from the ranking.

Build a list from your best customers, or read the guide to sizing a Swiss target market when you are starting from a written profile.