Website speed: what our test of 89 online shops showed

Only 3 of 89 top-rated Czech online shops loaded main mobile content within 2.5 s. See what slows them down, how to test speed in 5 minutes, and what to ask of your vendor.

Orange glass marble on a light ramp with small gray blocks in its path
WebsitesReading time: 9 min

Mobile website speed at top-rated Czech online shops lags behind Google's standards. On 14 November 2026, we used Google Lighthouse 13.6 to test the homepages of 89 finalists of Heureka ShopRoku 2025—an annual award by the Czech price-comparison site Heureka, decided by shoppers' votes and an expert jury. Main content loaded within 2.5 s on only 3 of them, while 76 took over 4 s.

Loading speed decides whether a shopper buys. Content matters more to Google—77 of the 89 shops scored 90 or above on SEO. Below, we share our findings, explain Core Web Vitals plainly, walk through a five-minute test, and outline how to speed up your store.

Why website speed matters: customers, not just Google

A slow page loses shoppers before they see your offer. On Google, speed acts mostly as a tiebreaker between similar content—lost orders hurt an online store far more.

Published Core Web Vitals case studies back this up. Vodafone in Italy (March 2021) ran an A/B test where a 31% better LCP brought 8% more sales. Japan's Rakuten 24 (web.dev study, 2022) lifted conversion rate by 33.13% and revenue per visitor by 53.37%. Enterprise results do not copy over one-to-one, yet the direction is clear.

According to Google Search Central (September 2026), Google uses Core Web Vitals, yet always prioritizes the most relevant content; green numbers in Search Console do not guarantee first place.

Google evaluates mobile and desktop separately at the 75th percentile (three-quarters of visits). Phones run slower due to weaker processors and mobile network latency—your site feels different on a tram than in the office.

What we found across 89 top-rated Czech online shops

We ran our test on 14 November 2026 in Google Lighthouse 13.6 (emulating a mid-tier phone on a slower network), measuring each homepage once. From 154 finalists and winners of Heureka ShopRoku 2025 (announced March 2026, excluding age-restricted goods and one mobile carrier), we randomly drew 90. Lighthouse failed to open one page, leaving 89.

Chart of mobile main content loading across 89 online shops: 3 within 2.5 s, 10 in 2.5–4 s, 76 over 4 s
Main content (LCP) on mobile across 89 finalists of Heureka ShopRoku 2025. Google Lighthouse 13.6 (mobile emulation), 14 November 2026, anonymized.

Only 3 shops finished within 2.5 s, 10 landed between 2.5 and 4 s, 76 exceeded 4 s, and 21 exceeded 10 s. The middle half sat between 5.6 and 9.8 s, with medians across five retail categories (electronics to garden) ranging from 7.2 to 8.0 s. First content (logo, menu) appeared at 2.2 s, main content at 7.8 s.

The median performance score (0–100) was 53: none reached 90, 4 scored 80 or above, and 34 fell below 50. Meanwhile, 77 shops scored 90 or above on SEO (meta descriptions, links, robots.txt)—and 66 of those took over 4 s to load main content.

The table summarizes the most common Lighthouse findings.

FindingShops of 89Impact on shoppers
Main content after 4 s76Shoppers wait past Google's limit for photos and offers
Render-blocking files82Screen waits until styles and scripts finish
Unused JavaScript72Phone downloads code it does not need
Short cache lifetime67Repeat visits re-download the same files
Oversized images60Photos exceed what the screen needs
Main thread blocked over 600 ms49Visible page briefly ignores taps
Text waiting on fonts35Headings and prices wait for font files

Cutting unused JavaScript would save the most time (34 shops), followed by images (21) and caching (19). Visual stability (CLS up to 0.1) passed on 83 of 89 shops. The median homepage weighed 3.2 MB across 142 requests.

What the numbers mean—and what they do not

This was a lab test of a first visit with an empty cache (150 ms latency, 1.6 Mbps—roughly the slowest quarter of 4G), not real visitors; faster connections show content sooner. Top-rated shops do not represent the market average, and we tested homepages once—the data highlights common bottlenecks, not store rankings.

According to HTTP Archive Web Almanac 2025, 48% of mobile sites and 45% of homepages passed Core Web Vitals in July 2025. That reflects real Chrome users, so it cannot be compared directly with lab runs.

LCP, INP, and CLS in plain English: how Google measures speed

Core Web Vitals are three Google metrics: LCP (loading), INP (responsiveness), and CLS (visual stability). Per Web Vitals, passing means hitting up to 2.5 s, 200 ms, and 0.1 on at least 75% of visits, separately on mobile and desktop.

MetricWhat it measuresGood thresholdStorefront analogy
LCPMain content displayup to 2.5 sWhen the shop window lights up
INPResponse to a tapup to 200 msHow fast a clerk replies
CLSWhether layout jumpsup to 0.1Whether the shelf stays still

LCP times the largest element in the first viewport, usually a hero banner or product photo (over 4 s is poor). INP measures response to taps like filters or Add to cart (over 500 ms is poor). CLS totals unexpected layout shifts, such as a late banner (over 0.25 is poor).

Field data vs. lab data

Field data comes from real visitors; lab data is a single simulated load. PageSpeed Insights pulls field metrics from the Chrome UX Report (CrUX) over the past 28 days. Google ranks using field data, while lab tests diagnose causes.

Numbers differ across devices, networks, and caches. Lab runs cannot measure INP because nobody taps—they show a proxy, Total Blocking Time (TBT).

The performance score is not Core Web Vitals

The 0–100 score in PageSpeed Insights is a lab grade, not Core Web Vitals. Lighthouse calculates it as a weighted average of five metrics, led by Total Blocking Time (30%), LCP (25%), and CLS (25%). Scores of 90+ are good, 50–89 need improvement, and under 50 are poor.

Where online shops lose time

In our test of Czech shops, time leaked through images, unused JavaScript, fonts, stylesheets, and short cache lifetimes. You can fix all of these without rebuilding your site.

Images: oversized, outdated, lazy-loaded

According to Web Almanac 2025, an image is the LCP element on 76% of mobile pages. Lighthouse flagged images on 60 shops: full-resolution uploads or JPG and PNG instead of WebP and AVIF.

Lazy loading on the top banner slows display—yet 16 to 17% of pages do it. Across 38 shops, the browser discovered the hero image late, failed to prioritize it, or lazy-loaded it.

Unused JavaScript

Unused JavaScript appeared on 72 of 89 shops and was the top bottleneck on 34. Phones needlessly download chat widgets, analytics tags, ad pixels, plugins, and unused theme code.

They block the page like a courier unpacking in the shop doorway. Total Blocking Time exceeded 600 ms on 49 shops (median 665 ms).

Fonts and styles delaying text

Text waited for custom fonts on 35 shops, while styles or scripts blocked rendering on 82. Fewer font weights, system fonts, and immediate fallback text solve this.

Caching and repeat visits

Browser cache stores site files on the visitor's phone. Cache lifetimes were too short on 67 shops, forcing returning shoppers to re-download logos, fonts, and scripts. Proper caching noticeably improves real-world LCP.

Read more on what a homepage should include and why shoppers abandon their cart.

How to test your website speed in 5 minutes

Test your site free in five minutes with PageSpeed Insights (individual pages) and Search Console (whole site):

  1. Test your homepage, a product page, and a category page on mobile in PageSpeed Insights.
  2. Check field data first: it shows 28-day Core Web Vitals (LCP, INP, CLS), though low-traffic Chrome sites may lack it.
  3. Review lab LCP and Total Blocking Time, then note the top three time savings.
  4. Open Core Web Vitals in Search Console to see page groups rated Good, Needs improvement, or Poor.
  5. Open your site on a budget phone over mobile data, watch how fast photos and prices appear, and tap the menu or filters.

Lab results fluctuate with server, network, and computer load. The median of five runs is twice as stable as a single run—always compare medians under identical settings.

What to fix first: from quick wins to a rebuild

Start with high-impact fixes: images, third-party scripts, fonts, and caching. JavaScript, images, and cache yielded the biggest savings in our test. Leave theme or platform changes for last.

Compress your homepage hero image into WebP or AVIF and disable lazy loading on it so it downloads first. Most themes let you set this directly.

Audit third-party scripts and remove dead weight like old tracking tags or unused chat tools. Defer the rest until after main content, cookie consent, or user interaction.

Have your developer trim font weights, show fallback text immediately, and extend cache lifetimes. If your setup hits hard limits, see how to choose an e-commerce platform or how to handle a website redesign without losing traffic.

Performance budget: what to ask of your agency

A performance budget puts speed limits into your contract. Think of it like a van's payload limit: adding a heavy feature means dropping another. Without one, sites slow down and nobody owns the problem.

Include mobile Core Web Vitals, pre-launch lab checks, and a page-weight cap in your brief (our sample's median was 3.2 MB).

Building speed in from the design stage costs the least. Planning an online store? Explore our e-commerce development or scope your project in our project configurator.

Frequently asked questions

How fast should a website load?

Focus on when main content appears. On mobile, aim for LCP within 2.5 s for at least 75% of visits (over 4 s is poor), responsiveness under 200 ms, and a stable layout. Among the 89 Czech online shops we tested, only 3 met the 2.5 s mark on their homepage.

Does website speed affect Google rankings?

Yes, though less than often claimed. Google factors in Core Web Vitals, yet prioritizes the most relevant content. Speed breaks ties between comparable pages. Its bigger impact is on shoppers—faster sites lift conversions and sales.

Why does PageSpeed Insights show a different score each time?

Lab tests vary with server, network, and device load. Run multiple tests and track the median—Lighthouse notes a five-run median is twice as stable. Field data in PageSpeed Insights aggregates 28 days, so it shifts slowly.

What are Core Web Vitals?

They are three Google metrics for real-user experience: LCP (main content, up to 2.5 s), INP (responsiveness, up to 200 ms), and CLS (visual stability, up to 0.1). Passing requires meeting them on at least 75% of visits. Check yours in PageSpeed Insights and Search Console.

Can you speed up an online store on a hosted platform?

Usually yes, within limits. You can compress images, delete unused apps and tracking codes, simplify your homepage, and pick a lighter theme. The provider controls the server, caching, and core scripts. If Core Web Vitals still fail, switching themes, platforms, or building custom is the next step.

More articles

Handpicked topics for you

Homepage design: what must be clear in 10 seconds

Visitors form a first impression before reading a word and decide within seconds whether to stay. Here is what must be clear at the top and how to test it in an afternoon.

Reading time: 9 min

Cart abandonment: 10 reasons shoppers leave checkout

Seven in ten carts end without an order. Some shoppers just compare prices, while shipping costs, forced accounts, or missing parcel lockers stop others. Here is what to fix.

Reading time: 10 min

Choosing an e-commerce platform: Shopify, Shoptet or custom

Most new online stores only need a hosted platform. Learn how to spot when you outgrow it and when custom e-commerce is an unnecessary expense.

Reading time: 9 min