Marketing and brandingwebsite speed in French-speaking Switzerland

Website speed in French-speaking Switzerland: 1'820 sites analysed, a developing country, digitally speaking?

Our PageSpeed survey of 1'820 sites in French-speaking Switzerland: what it measures, its limits, and the real link between speed, Core Web Vitals and Google.

Mehdi B.

Co-founder · SEO & digital strategy

Website speed in French-speaking Switzerland: 1'820 sites analysed, a developing country, digitally speaking?
In brief

Of 1'820 French-speaking Swiss company websites tested in December 2025 with PageSpeed Insights on the mobile profile, 7.4% score 90 or more. This lab score is not a ranking criterion: Google relies on Core Web Vitals measured on real visitors, among other signals. We recommend optimising what exists, images and scripts first, rather than rebuilding everything.

One Sunday evening, the manager of an SME in Lausanne runs PageSpeed Insights on his website, sees a 54 on an orange background and wonders whether he has just lost his place on Google. The answer is no. This score is a lab diagnostic, very useful for understanding what slows a page down, but it is not the criterion Google uses to rank sites. To measure website speed in French-speaking Switzerland, we ran 1’820 company websites from the region through this same tool in December 2025, and the result is harsh: only 7.4% get a green score.

This figure deserves to be read with its limits. Our view comes down to a few words. Most of these sites do not need a redesign. They need targeted work on images and scripts, then a check on real visitors, which we do as part of our digital marketing support.

What our survey of 1’820 websites measures, and what it does not

The survey was carried out in December 2025 using the Google PageSpeed Insights API, on the mobile profile. The sample comprises 1’820 unique websites of companies based in French-speaking Switzerland, from small local SMEs to large institutions. We chose the mobile profile because it is the one that counts for the search engine: according to Google's documentation, it is the mobile version of a site, crawled by the smartphone bot, that is used for indexing and ranking.

This mobile profile is deliberately demanding. According to the PageSpeed Insights documentation, the test simulates a mid-range device, and the documentation for Lighthouse, the tool's analysis engine, specifies the network conditions applied: 150 milliseconds of latency and 1.6 Mbit/s of download speed, which roughly corresponds to the slowest quarter of 4G connections. The processor is also slowed down fourfold. In other words, the score describes the experience of a visitor poorly served by their network, not that of an employee on fibre at the office.

We must also say what this survey does not measure. It is a lab score, obtained on a given date, on one page per site. It says nothing about the other pages, nor about what visitors actually experience. It also varies from one run to the next: the Lighthouse documentation acknowledges that the score fluctuates with test conditions, the network or the advertising content loaded, and recommends reading it as a distribution rather than a single figure. Finally, we do not present these 1’820 sites as a representative sample in the statistical sense. It is a broad snapshot of the web in French-speaking Switzerland, not a poll.

Website speed in French-speaking Switzerland in figures

PageSpeed Insights sorts scores into three zones: 90 and above for a score judged good, 50 to 89 for a score needing improvement, below 50 for a poor score. Here is how the 1’820 sites in the survey are distributed:

  • Green zone, 7.4% of sites: a score of 90 or more on mobile.
  • Orange zone, 76.3% of sites: a score between 50 and 89.
  • Red zone, 16.4% of sites: a score below 50.

Three other observations emerge from the technical data collected during the survey. First, the average weight of a tested page is around 4 MB, of which 2.1 MB is images on average. Second, on most sites, the display of the page's main element exceeds, in the lab, the 2.5-second threshold recommended by Google. Finally, we observed a gap of 19 score points between sites that load little JavaScript and those that load a lot.

What these figures allow us to state is fairly simple: on an average phone and a slow network, the vast majority of the French-speaking Swiss sites tested have room for improvement. What they do not allow us to state matters just as much. A score of 70 does not prove that the site is losing customers, or that it is slipping in Google. It points to technical items to examine, nothing more.

Why a score of 90 remains rare, in Switzerland as elsewhere

Before concluding that French-speaking Switzerland is lagging behind, you need to understand how the score is built. According to the Lighthouse documentation, it combines five weighted measurements: processor blocking time accounts for 30%, display of the main element 25%, layout stability 25%, and first paint and speed index 10% each.

Each measurement is then converted into a score based on real site data observed by HTTP Archive. The scale is severe by design: a value that only about 8% of the sites in that dataset match or beat earns 90 points. A green score therefore rewards top-tier performance, and it is normal for only a small minority of sites to achieve it. We do not have a global distribution of PageSpeed scores that would let us say whether French-speaking Switzerland does better or worse than its neighbours. The 7.4% in the survey describes the rarity of technical excellence, not underdevelopment.

The weighting also explains the 19-point gap linked to JavaScript. Blocking time, which measures how long the phone's processor stays busy with scripts, accounts on its own for 30% of the score. A site loaded with third-party tools is penalised mechanically on this item.

What Google really evaluates: Core Web Vitals measured on your visitors

Google does not rank pages according to their PageSpeed score. It relies in particular on Core Web Vitals, three experience metrics measured on real Chrome users and gathered in the Chrome User Experience Report, known as CrUX. Each has a threshold defined by Google:

  • LCP (Largest Contentful Paint): the time taken to display the largest visible element, an image or a block of text. It should stay at 2.5 seconds or less.
  • INP (Interaction to Next Paint): the delay between a click or input and the page's visible reaction. It should stay at 200 milliseconds or less. This metric replaced the former FID in March 2024.
  • CLS (Cumulative Layout Shift): the stability of the page during loading, that is, elements that move under the visitor's finger. The score should stay at 0.1 or less.

These thresholds are assessed at the 75th percentile of visits, separately on mobile and desktop, as Google's web.dev documentation specifies. In other words, three visits out of four must pass the bar. The lab cannot measure INP anyway, for lack of a real visitor who clicks: Lighthouse replaces it with processor blocking time, which is only an approximation.

On the role of these metrics in SEO, the Google Search Central documentation is clear. Core Web Vitals are used by ranking systems, but there is no single page experience signal. Good results do not guarantee a top position in the results, and Google always seeks to show the most relevant content, even when the page experience is poor. A slow site is therefore not demoted mechanically. With comparable relevance, page experience can play a part, but the main reason for speed remains the visitor waiting in front of their screen.

The right tool to track these metrics is the Core Web Vitals report in Google Search Console. It is based on real CrUX data, groups pages that offer a similar experience and gives each group the status of its weakest metric. A group whose LCP is good but whose CLS is poor is therefore classed as poor.

The survey against external studies: Web Almanac, CrUX, KMU Digital Pulse

HTTP Archive's Web Almanac 2025, published in January 2026, offers a solid point of comparison. Its performance chapter draws on CrUX data from July 2025, covering millions of sites worldwide. It shows that 48% of sites have good Core Web Vitals on mobile, against 56% on desktop. The progress is clear, since the mobile share was only 32% in 2021 and 44% in 2024. Metric by metric, on mobile, 62% of sites have a good LCP, 77% a good INP and 81% a good CLS. Displaying the main element therefore remains the weak point of the global web, as in our survey.

The same chapter reports a median blocking time of 1’916 milliseconds on mobile, up 58% over a year. Scripts are getting heavier, and not only in French-speaking Switzerland. The chapter on page weight, for its part, gives a median mobile home page of 2’559 KB, of which 911 KB is images, in July 2025.

These figures cannot be compared directly with ours. The Web Almanac's 48% is field data, whereas our 7.4% is lab scores: a site can pass its Core Web Vitals with its visitors and still have a PageSpeed score of 75. Likewise, our 4 MB is an average, pulled upwards by a few very heavy pages, whereas the Web Almanac publishes medians. The comparison suggests, at most, that the pages tested in French-speaking Switzerland are no lighter than the global average.

There is one limit specific to SMEs. CrUX includes only public sites with enough traffic to produce statistically reliable data. Many French-speaking Swiss SME websites do not reach this threshold, so PageSpeed Insights and Search Console show no field data for them. In that case, the lab score remains the only reference available, provided you repeat the test and apply it to several types of page.

Swiss studies, finally, do not measure speed, but they put its importance in perspective. The KMU Digital Pulse 2025, conducted by localsearch and the Lucerne University of Applied Sciences and Arts (HSLU), combines an online survey of 1’660 consumers, carried out from 23 June to 2 July 2025, and an analysis of 47’079 Swiss SME websites. According to the study, 80% of people looking for an SME they do not know start with a search engine. It also finds that 77% of the population would like to be able to book an appointment online, while only 3% of SMEs offer this option on their website. Its scope is national and does not single out French-speaking Switzerland, but the message is clear: even before speed, what is often missing is what the site lets people do.

Where to start on an existing site

Take a ten-person accounting firm in Fribourg, whose website scores 58 on mobile. The temptation would be to rebuild everything. Our approach is the opposite: it starts with measurement, then tackles the causes in the order in which they weigh.

The first step is to check whether field data exists. PageSpeed Insights shows it when available, over a 28-day period, and the Search Console Core Web Vitals report breaks it down by group of pages. If nothing appears, the site does not have enough traffic to feature in CrUX, and we then work from several lab tests on the home page, a service page and a contact page.

Next come the images, which often explain a slow LCP. Google's guide to LCP optimisation recommends modern formats such as AVIF or WebP, a main image discovered directly in the page's code, and above all never deferring the loading of that main image. In practice, a header photo of several megabytes, exported as is from a photo shoot, is replaced with a version sized for the visitor's screen.

The third area concerns third-party scripts: chat windows, advertising pixels, social media modules, testing tools. Each one occupies the phone's processor, which worsens blocking time in the lab and INP for visitors. We draw up an inventory with the client, remove those that are no longer used and defer those that are not needed for the first display.

The last point concerns the server and the cache. According to the same Google guide, server response time should account for around 40% of the LCP, and a content delivery network brings files closer to the visitor. We detail this layer in our article on our hosting setup for Elementor websites.

Once these fixes are live, you need to wait a few weeks before judging, since field data covers a rolling 28 days. This monitoring promises no position in Google. It does, however, tell you, with figures, whether the site works better for those who visit it from a train between Lausanne and Geneva or from a building site in Valais.

If you would like to know which zone your site falls into and where to start, that is precisely the purpose of the first conversation: get in touch with us.

Sources

web performanceCore Web VitalsSEO
Mehdi B.
About the authorMehdi B.

Co-founder of Smart Impact. Hooked on the web from the start, he launched his first project in 2006: an online music magazine still running today. With close to 20 years of SEO experience, a Swiss federal diploma in marketing and a solid geek streak, he turns clients' ideas, sometimes hazy ones, into concrete digital projects with his team.

LinkedIn
Further reading