Nexilon

Nexilon field study  ·  1,883 Swiss SME websites  ·  16 February – 24 August 2026

Swiss SMEs bought design. Nobody sold them the finishing.

Between 16 February and 24 August 2026 we opened 1,883 Swiss company websites, one at a time, in a real browser. The average site scored 81.8 out of 100 for quality — and 50.0 out of 100 for legal compliance. That gap is not a design problem. It is the invoice nobody sent.

Average website quality81.8
Average legal compliance50.0
31.8 points between how a Swiss website looks
and whether it holds up · n = 1,883
Check my site — free

The same 20 checkpoints we ran on 1,883 others. Your result in writing, within two working days.

Finding 01 What we did not find

The average Swiss SME website is beautiful and half-legal

We went looking for bad websites. We found the opposite. Of the sites we could reach, 1,789 out of 1,790 serve HTTPS correctly. Only 2.4% fail a basic mobile check. Somebody paid for these sites, and in most cases somebody good built them.

Then we looked below the fold.

One company in four publishes no Impressum. One in five publishes no Datenschutzerklärung. And 232 companies — 12.3% of everything we opened — publish neither.

That is not carelessness. It is a scope boundary. Your designer was paid to make the site look right. Nobody in that transaction was paid to make it lawful.

26.4%
No Impressum
498 of 1,883
19.2%
No Datenschutz­erklärung
362 of 1,883
12.3%
Neither of the two
232 of 1,883
2.4%
Fail on mobile — this
is not your problem

Finding 02 Consent and tracking

One in four Swiss websites watches visitors without asking

We fingerprinted tracking scripts on 987 sites. We found a consent layer on 797. The difference — 471 companies — runs analytics, advertising pixels or session recording with nothing standing in front of it.

Since 1 September 2023 the revised Swiss Data Protection Act has required you to tell people what personal data you collect and why (nDSG Art. 19). UWG Art. 3 Abs. 1 lit. s has required a commercial website to say who is behind it for far longer. Neither obligation is new. Both are widely unfinished.

And the gap is widest in exactly the wrong places: veterinary practices at 28.9%, mental-health practices at 26.9%, health and wellness at 26.2%. The businesses handling the most sensitive personal data in this study are the least prepared to answer a question about it.

25 of every 100 sites we opened. 47.7% of every Swiss SME site that tracks at all has nothing in front of it. Not a banner that lets everything through — no banner at all.

Finding 03 Where customers actually look

A third of these businesses cannot be found where their customers search

641 companies — 34.0% — have no Google Business Profile at all. Among those that do, the median review count is six, and 342 publish no opening hours, so a customer standing on the pavement at 17:40 cannot tell whether it is worth walking in.

Then we measured something nobody had asked us to measure: the relationship between how good a company's website is and how well it does in local search.

The correlation is +0.046. Statistically, that is nothing at all.

Read that twice, because it contradicts what almost every owner believes. A better website does not make you findable. These are two independent systems with two separate fixes — and almost nobody sells them separately, which is why almost nobody fixes both.

34.0%
No Google Business
Profile · 641 of 1,883
6
Median reviews where
a profile does exist
342
Profiles publishing
no opening hours
+0.046
Correlation: site quality
× local rating

Method Why our number differs from everybody else's

One site in twelve is telling its owner a lie

Any tool can request a URL and check whether /impressum returns a 200. We built one that does not trust the answer.

Modern platforms route unknown addresses to a catch-all page. Ask such a site for /impressum and it returns a page. Ask it for an address that was never created and it returns the same page. The owner runs a checker, sees green, and stops thinking about it. There is no Impressum. There never was.

So before reading any legal page on any site, we asked for a page we knew did not exist. 155 sites — 8.2% — failed that probe. Every one of them reads as compliant to a cheaper tool.

Negative-control probe · executed before any legal page is read
GET /diese-seite-existiert-nicht-8f3a  →  HTTP 200, 4.1 kB of contentCATCH-ALL
GET /impressum  →  HTTP 200, identical contentPHANTOM PAGE
Verdict recorded for this siteIMPRESSUM ABSENT
155 of 1,883 sites (8.2%) behaved exactly like this.

We were not clever for thinking of this. We were burned into it. On our very first batch we went back and re-read by hand what the automation had reported: 10 of 33 readings were wrong. A 30% error rate.

From the second batch onward a person opened and read every doubtful page before a verdict was recorded. Across nineteen batches, hand-verification overturned between 19 and 36 readings in every hundred. That is the whole difference between a number you can quote and a number you cannot — and the reason this took nine days instead of an afternoon.

Read the full method — population, instrument, verification, and what we deliberately did not measure →

Position The sentence we will not say

Roughly one Swiss website in ten needs replacing. We will not pretend yours is one of them.

“Your website is bad” addresses one owner in ten and insults the other nine. We counted, so we know the ratio, and we do not open with it.

What almost everyone actually needs is finishing. A legal page that says what it must say. A consent layer that blocks something rather than merely appearing. A Google profile with opening hours on it. A path from interested to in touch. Each one bounded, specific, checkable — and a week's work at most.

That is a different conversation entirely. It starts with knowing which of the five you are missing.

Next step Free · no obligation · no call required

Find out which one you are

  • We run the same scan on your site that we ran on 1,883 others: 20 compliance checkpoints, your consent and tracking layer, your Google Business Profile, and the paths a visitor has to reach you.
  • You get the result in writing within two working days, with the checkpoint and the evidence for every line.
  • If everything is in order we tell you that and stop. Several hundred companies in this study needed nothing structural at all, and we would rather be the people who told you so.
  • We never publish company-level results. The method is public, the aggregate is public. Who scored what is not, and never will be.
Scroll to Top