Speed and AI readiness of Czech online stores

The lab says 12 seconds; people see 1.7

Mobile Lighthouse simulates a slower phone on a poorer network. In that simulation, the main content on the home page loads in a median of 12 seconds, and 94% of online stores are over 4 seconds. Real visitors, whose data Chrome collects, see the main content in a median of 1.7 seconds, and 90% of online stores are within the threshold. Google uses real-user data for search rankings. But the lab shows how the page performs for a customer with an older phone or a weak signal - and there are more of them than it seems from an office with fast Wi-Fi.

The real problem is responsiveness, not loading

58 of 84 online stores do not meet Core Web Vitals, and for 56 of them the reason is response to clicks (INP). A customer taps a filter, variant, or add-to-cart button and the page does not respond for a while. Responsiveness is mainly worsened by JavaScript blocking the browser's main thread: in the lab, Total Blocking Time measures this, with a median of 551 ms and 46% of online stores above 600 ms. Fewer third-party scripts, code splitting, and deferring everything not needed for the first interaction help.

AI crawlers have the door open, but often cannot see the content

Most large online stores do not block AI crawlers: 16% block at least one, while 10% block crawlers from AI search engines and assistants. But an open door is not enough. A crawler that does not run JavaScript sees only what is in the HTML response, and for the median online store that is 57% of the text a person sees. For 16% of online stores, it is less than a fifth. 24% of online stores have an llms.txt file with a short description for AI assistants, and 70% have structured data in their HTML.

What to take from this

These are the largest Czech online stores with their own development teams, yet their home pages do not meet Google's responsiveness thresholds on average. For a smaller online store, this is an opportunity: fast responsiveness and content readable without JavaScript are not a matter of budget, but of how the website is built. We measure the same metrics on your online store at no charge as part of a website audit.

Methodology

Sample size

85 online stores from the Top 100 Czech online stores ranking (CzechCrunch × BizMachine, June 2025) with a .cz or .eu domain. We excluded Mall.cz because it redirects to another online store in the sample. We measured the home page of each one.

Measurement method

Speed: Google PageSpeed Insights API (Lighthouse 13) - the mobile profile up to three times and the median of the runs, desktop once. Real-visitor data: Chrome UX Report (75th percentile over the last 28 days), as returned by PageSpeed Insights. AI readiness: rules for 10 AI crawlers in robots.txt, whether a /llms.txt file exists, and the share of visible text that is already in the HTML before JavaScript runs (comparing the HTML response with the rendered page in Chromium).

Period

24 October 2026, one-off measurement

Limitations

Home pages only - categories, products, and the cart may produce different results. The lab measurement simulates a slower phone on a mobile network, so it scores worse than real people's experience; we report both separately. PageSpeed Insights may return a cached result, so for some online stores the three requests returned the same measurement. Mobile measurement failed for 4 online stores, and robots.txt was retrieved for 80 of 85. The sample consists of the largest Czech online stores, not a typical Czech online store. We do not name individual online stores; we publish aggregate figures only.

Related

• Article: How fast are the biggest Czech online stores: measuring 85 shops
• Web design and development for online stores
• Online stores for second-hand marketplaces