Re-measured and rewritten on September 17, 2026
The April version of this study said only 1 of 48 DACH publishers passes Core Web Vitals on mobile. That was wrong, for two reasons. First, it used a single Lighthouse lab run per site, but Core Web Vitals are assessed on field data from real Chrome users. Second, the one site it counted as passing, puls24.com, was returning a Cloudflare error page at the time. The April text also explained the scores with tech-stack and ad-stack details that weren't actually measured. I re-ran all 50 sites with both lab and field data, and the result points the other way.
Most large news sites in Germany, Austria and Switzerland pass Core Web Vitals. On September 17, 2026, 36 of the 47 sites I could measure met all three thresholds on mobile, based on Chrome UX Report data from real users. Their PageSpeed lab scores tell a very different story: the median mobile score is 52, and only one site reaches 90.
This is the third study in the DACH media series. Study 1 looked at which AI crawlers publishers block in robots.txt. Study 2 measured how much content only appears after JavaScript runs. This one looks at how fast the pages are, and at why the two ways of measuring that disagree so much.
The numbers: 36 of 47 pass
47 DACH news sites, lab and field data.
PageSpeed Insights, mobile and desktop, for 50 publishers. Three didn't return a working page to my script (puls24.com, zeit.de, tagesanzeiger.ch). Pass means the origin's 75th percentile meets all three Core Web Vitals: LCP ≤ 2.5 s, INP ≤ 200 ms, CLS ≤ 0.1.
The pattern is consistent: real users mostly get fast pages, while a throttled lab run of the same homepage looks slow. The sites that fail in the field don't fail on loading speed. None of them misses LCP on mobile. They fail on responsiveness (INP) or on layout stability (CLS).
Lab score vs field data: why they disagree
PageSpeed Insights shows two kinds of data on the same page, and they measure different things.
Field data comes from the Chrome UX Report (CrUX). It's collected from real Chrome users over the last 28 days and reported at the 75th percentile. For this study I used the origin-level numbers, which cover all pages of a site, not just the homepage. Google's Core Web Vitals assessment is based on this data.
Lab data comes from a single Lighthouse run. For mobile, Lighthouse emulates a mid-range phone with a throttled CPU and a slow 4G connection, and it loads one URL with an empty cache. The performance score (0 to 100) is calculated from that run only.
For news sites, those two situations are far apart. Many readers come back several times a day, so files are already cached. Many use recent phones on fast connections. A lab run gets none of that, and a throttled CPU spends a long time on third-party scripts before the largest element is painted. That's how br.de ends up with a 39.5-second LCP in the lab and 1.3 seconds for the 75th percentile of its actual visitors.
Neither number is wrong. The lab score is useful for finding what slows a page down and for comparing changes before they ship. But a lab score of 35 doesn't mean a site fails Core Web Vitals, and a score of 97 doesn't mean it passes. derstandard.at has the best mobile lab score in the sample and still fails in the field, because real users see too much layout shift.
Who fails, and on which metric
Eleven sites fail Core Web Vitals on mobile. None of them because of LCP.
- Six fail only on INP: zdf.de (260 ms), 20min.ch (445 ms), faz.net (244 ms), stern.de (214 ms), tt.com (204 ms) and merkur.de (203 ms). The last three are just over the 200 ms line.
- Three fail only on CLS: nzz.ch (0.59), derstandard.at (0.20) and taz.de (0.14).
- Two fail on both: kleinezeitung.at (INP 239 ms, CLS 0.13) and sn.at (INP 244 ms, CLS 0.11).
On desktop the picture shifts toward layout shift. 15 sites miss the CLS threshold there, including spiegel.de, handelsblatt.com and wdr.de, which reaches 0.90. INP is barely a problem on desktop: only 20min.ch misses it. Two sites miss desktop LCP: handelsblatt.com at 3.0 s and kleinezeitung.at at 4.0 s.
I can't tell you from this data why a specific site has high INP or CLS. PageSpeed Insights doesn't attribute field data to scripts or elements. For that, a publisher needs its own real-user monitoring or a closer look at its templates. What the data does show is where to look: on these sites it's interaction and layout, not first load.
Germany, Austria, Switzerland compared
| Country | Sites measured | Pass CWV (mobile) | Median lab score (mobile) |
|---|---|---|---|
| Germany | 24 | 19 | 53 |
| Austria | 17 | 13 | 50 |
| Switzerland | 6 | 4 | 36 |
The pass rates are close: 79% in Germany, 76% in Austria, 67% in Switzerland. With only six Swiss sites measured, that gap doesn't mean much. By type of publication, all five business titles and all three tech sites pass on mobile, and six of seven public broadcasters do. Regional papers do worst: only one of four passes.
All 47 sites
Mobile values, sorted by lab score. LCP, INP and CLS are the origin's 75th percentile from CrUX.
| Site | Lab | LCP | INP | CLS | CWV |
|---|---|---|---|---|---|
| derstandard.at | 97 | 1.0 s | 106 ms | 0.20 | ✗ |
| wienerzeitung.at | 83 | 1.1 s | 121 ms | 0.00 | ✓ |
| golem.de | 80 | 1.1 s | 119 ms | 0.03 | ✓ |
| heise.de | 77 | 1.6 s | 81 ms | 0.00 | ✓ |
| welt.de | 73 | 1.5 s | 195 ms | 0.05 | ✓ |
| wdr.de | 73 | 1.6 s | 101 ms | 0.00 | ✓ |
| wiwo.de | 73 | 1.4 s | 137 ms | 0.04 | ✓ |
| ndr.de | 67 | 1.3 s | 80 ms | 0.00 | ✓ |
| zdf.de | 66 | 1.8 s | 260 ms | 0.03 | ✗ |
| taz.de | 64 | 1.2 s | 104 ms | 0.14 | ✗ |
| heute.at | 62 | 1.0 s | 138 ms | 0.07 | ✓ |
| manager-magazin.de | 62 | 1.1 s | 128 ms | 0.01 | ✓ |
| t3n.de | 62 | 1.2 s | 116 ms | 0.07 | ✓ |
| watson.ch | 61 | 1.2 s | 118 ms | 0.10 | ✓ |
| srf.ch | 60 | 1.4 s | 113 ms | 0.04 | ✓ |
| nachrichten.at | 59 | 1.5 s | 139 ms | 0.01 | ✓ |
| focus.de | 58 | 1.2 s | 137 ms | 0.07 | ✓ |
| diepresse.com | 55 | 1.3 s | 112 ms | 0.02 | ✓ |
| news.at | 55 | 1.9 s | 96 ms | 0.01 | ✓ |
| orf.at | 55 | 0.8 s | 145 ms | 0.00 | ✓ |
| spiegel.de | 53 | 1.0 s | 133 ms | 0.01 | ✓ |
| handelsblatt.com | 53 | 1.2 s | 182 ms | 0.04 | ✓ |
| capital.de | 53 | 1.7 s | 164 ms | 0.03 | ✓ |
| oe24.at | 52 | 0.8 s | 90 ms | 0.03 | ✓ |
| merkur.de | 51 | 1.0 s | 203 ms | 0.06 | ✗ |
| kurier.at | 50 | 1.2 s | 163 ms | 0.04 | ✓ |
| tagesschau.de | 50 | 1.1 s | 95 ms | 0.02 | ✓ |
| profil.at | 48 | 1.5 s | 111 ms | 0.10 | ✓ |
| stern.de | 48 | 1.7 s | 214 ms | 0.03 | ✗ |
| fr.de | 48 | 1.0 s | 181 ms | 0.07 | ✓ |
| bild.de | 47 | 0.9 s | 136 ms | 0.04 | ✓ |
| br.de | 47 | 1.3 s | 96 ms | 0.09 | ✓ |
| trend.at | 44 | 1.3 s | 104 ms | 0.00 | ✓ |
| kleinezeitung.at | 38 | 2.4 s | 239 ms | 0.13 | ✗ |
| sn.at | 38 | 1.4 s | 244 ms | 0.11 | ✗ |
| blick.ch | 38 | 1.3 s | 140 ms | 0.08 | ✓ |
| krone.at | 35 | 1.4 s | 166 ms | 0.05 | ✓ |
| sueddeutsche.de | 34 | 1.4 s | 162 ms | 0.05 | ✓ |
| 20min.ch | 34 | 1.9 s | 445 ms | 0.08 | ✗ |
| falter.at | 33 | 1.6 s | 102 ms | 0.10 | ✓ |
| servustv.com | 33 | 1.9 s | 192 ms | 0.04 | ✓ |
| n-tv.de | 28 | 1.2 s | 167 ms | 0.05 | ✓ |
| tt.com | 27 | 2.0 s | 204 ms | 0.08 | ✗ |
| beobachter.ch | 25 | 1.4 s | 118 ms | 0.09 | ✓ |
| faz.net | 23 | 1.6 s | 244 ms | 0.06 | ✗ |
| tagesspiegel.de | 21 | 1.5 s | 168 ms | 0.10 | ✓ |
| nzz.ch | 9 | 1.8 s | 117 ms | 0.59 | ✗ |
What publishers can actually do
If most sites already pass, the useful question is how to keep passing and where the remaining risk sits.
- Check field data first. The Core Web Vitals report in Search Console and the field section at the top of PageSpeed Insights show what Google assesses. Don't start a performance project because of a lab score alone.
- Look at INP on article pages. Six of the eleven failing sites miss only INP. Long tasks during interactions come from event handlers, consent dialogs, ad refreshes and heavy scripts on the main thread. Chrome DevTools' Performance panel shows which tasks block an interaction.
- Reserve space for anything that loads late. Ads, embeds, cookie banners and "read more" boxes that push content down cause CLS. Fixed slot sizes and
min-heighton ad containers are the standard fix. - Test desktop separately. 15 sites miss CLS on desktop but only 5 on mobile. Wider layouts with more ad slots behave differently.
- Use lab runs to find causes, not to grade. A Lighthouse run shows render-blocking resources, main-thread work and the LCP element. That's valuable for debugging, even if the score itself doesn't decide the assessment.
Does it matter for rankings? Google says its ranking systems use Core Web Vitals as part of page experience, and that good page experience alone doesn't guarantee top rankings. For a news site, the stronger reason is readers: a page that jumps while someone tries to tap a headline is annoying regardless of what Google does with it.
Methodology
The list is the same 50 publishers used in Study 1 and Study 2: 25 from Germany, 18 from Austria and 7 from Switzerland, covering dailies, tabloids, regional papers, magazines, public broadcasters, business and tech titles.
On September 17, 2026, a script first requested each homepage with a normal browser user agent and followed redirects. Only sites that answered with HTTP 200 were tested, so an error page can't be counted as a result. Three were excluded: puls24.com (no connection), zeit.de (403) and tagesanzeiger.ch (no 200 response). For the remaining 47, I called the PageSpeed Insights API through Lumina's worker for both mobile and desktop, Lighthouse 13.4.1, category performance.
From each response I stored the lab metrics (score, FCP, LCP, TBT, CLS, Speed Index) and the origin field data (originLoadingExperience): the 75th percentile and category for LCP, INP, CLS, FCP and TTFB. All 47 sites had origin field data for both strategies. A site passes when LCP ≤ 2,500 ms, INP ≤ 200 ms and CLS ≤ 0.1 at the 75th percentile. That matches PageSpeed Insights' own overall category: exactly the 36 passing sites are rated FAST on mobile.
Each lab value is a single run and varies from run to run. The field data is a 28-day rolling window, so it changes slowly. The raw PSI responses and the analysis script are stored with the study, so every number here can be recomputed.
FAQ
Where to start
Five steps for a publisher that wants to know where it stands.
Run your homepage through Lumina's PageSpeed tool or PSI and look at the field section first: LCP, INP and CLS at the 75th percentile, for the URL and for the whole origin.
PageSpeed Tool →The Core Web Vitals report in Search Console groups similar URLs, so you can see whether article pages, galleries or live blogs are the problem.
Core Web Vitals Guide →Record an interaction in Chrome's Performance panel, such as opening the menu or accepting consent, and look for long tasks. That's where INP comes from.
Check main-thread work →Reserve fixed space for ads, embeds and banners. 15 of 47 sites in this study miss CLS on desktop, three times as many as on mobile.
Check desktop CLS →Field data moves on a 28-day window. A monthly check shows whether a change helped real users, and catches regressions from new ad partners or scripts.
PageSpeed Tool →Compare your site with the DACH sample
Lumina's free PageSpeed tool runs the same PageSpeed Insights test I used here and shows field data next to the lab score. One URL, no signup.
Run the PageSpeed Tool →