How fast do SME websites in Zurich, Winterthur, Bern and Basel load?

Key findings

• In the mobile lab test (simulated slow 4G), the median homepage took 7.6 seconds until its largest content was visible (Largest Contentful Paint, LCP). For a quarter of the sites it took more than 12.3 seconds.
• Only 23 percent of the sites reach 90 points or more in Lighthouse. The median is 71 out of 100 points.
• The median homepage weighs 2.0 MB and loads 54 files.
• The most common slowdowns: unused JavaScript (42 percent of sites) and images not in WebP or AVIF (38 percent).
• Sites from Winterthur were the fastest (median 5.2 seconds), sites from Zurich the slowest (8.4 seconds).

Results by city

The differences between the cities are small compared with the differences between individual websites. Whether a site loads fast depends on how it is built and hosted, not on where the business is.

What slows the sites down most

For every page, Lighthouse estimates how much sooner the largest content would appear with a given fix. We counted a slowdown when the estimated saving was at least half a second.

Then there is page weight: 23 percent of the homepages are larger than 5 MB.

Website builder, WordPress and others: does the system make a difference?

The system alone makes little difference. A lean WordPress site can be faster than an overloaded site builder page, and vice versa. What matters is image sizes, the number of scripts and the hosting.

What to check on your website first

• Images: upload photos at the size they are shown, in WebP or AVIF. This is often the biggest and easiest saving.
• Scripts and plugins: every chat widget, slider and tracking tool loads its own files. Remove what nobody needs.
• Hosting and caching: if the server responds slowly, no on-page optimization will help. A page cache and good hosting come first.
• Measure instead of guessing: test your homepage with PageSpeed Insights. It also shows data from real visitors if your site gets enough traffic.

If you want to know what slows your site down, we will look at it with you on a first call. Read more about our work on our Web development and Get a website built pages.

How we measured

• Sample: 289 business websites from Zurich, Winterthur, Bern and Basel, drawn at random from all OpenStreetMap entries with a website. They include shops, restaurants, medical practices, trades, offices and service providers. We excluded chains, authorities, public institutions, museums, associations and large companies, first automatically and then by hand. Each domain appears only once.
• We analyzed 217 websites. Of the rest, 36 were unreachable or did not return a web page, and we excluded 24 by hand during review. Another 8 ran on a subdomain of a larger organization or a website builder, and for 4 the measurement failed.
• Tool: Google Lighthouse 12.8.2, mobile profile with simulated slow 4G, one run per website, homepage only. Measured on 27 September 2026 from Vilnius.
• Limits: these are lab measurements. Real visitors on fast Wi-Fi often see the pages load faster. We measured each website once, and a single run can vary by several seconds. The value for one website is therefore uncertain, while the medians across many websites are much more stable.
• We do not name individual businesses. That is why the raw data contains no business names or web addresses.

Download the data

All measurements as a CSV file, without names or domains: rekos-speed-studie-2026-daten.csv (column names in German). You may use the data freely with the credit “Rekos Speed Study 2026” (license CC BY 4.0). The sample comes from OpenStreetMap, © OpenStreetMap contributors, license ODbL.