Core Web Vitals: Why Fast Sites Win Customers

Somewhere in Google’s index right now, your website has a report card. Three grades, updated continuously, based on what real visitors experience when they open your pages. Most business owners have never seen it. Your competitors’ agencies have. And if your grades are bad, that’s an opportunity someone else is already using.

The report card is called Core Web Vitals. Here’s what it measures, why it moves both rankings and revenue, and how to check your own site before you finish your coffee.

The three metrics, in plain language

Google boiled “does this site feel fast?” down to three numbers. No developer vocabulary required:

Metric	The question it answers	Good	Poor
LCP (Largest Contentful Paint)	How long until the main content is visible?	≤ 2.5 s	> 4.0 s
INP (Interaction to Next Paint)	When I tap or click, how fast does the site react?	≤ 200 ms	> 500 ms
CLS (Cumulative Layout Shift)	Does the page jump around while loading?	≤ 0.1	> 0.25

Translated into customer experience:

• LCP is the restaurant guest waiting to be seated. At 2 seconds nobody notices. At 5 seconds on a mobile connection, they’re back on Google tapping the next result.
• INP is pressing a lift button and nothing lighting up. A menu that opens 600 ms after the tap makes the whole site feel broken, even if nothing is.
• CLS is reaching for “Order” and the button jumping because a banner loaded above it: the reason people accidentally click ads on cheaply built sites.

One important detail: Google grades these from field data, measurements collected from real Chrome users on your actual site over the previous 28 days. You can’t fake it for the test, and a fast office Wi-Fi connection doesn’t save you. Your customer on LTE between metro stations is the measurement.

What speed does to rankings, and to revenue

Core Web Vitals have been a Google ranking signal since 2021. Google is deliberately vague about the weight, and honesty requires saying so: vitals alone won’t outrank better content. They work as a tiebreaker, and in a local market like “accountant Prague” or “plumber Vinohrady”, where twenty businesses offer roughly the same thing, tiebreakers decide who’s on page one.

The revenue effect is better documented than the ranking effect, and it’s bigger:

• Google’s own mobile research found that as load time goes from 1 s to 3 s, the probability of a visitor bouncing rises by 32%; from 1 s to 5 s, by 90%.
• A Deloitte/Google study of retail and travel sites measured that a 0.1-second speed improvement lifted conversion rates by roughly 8% in retail.
• Sites that pass all three Core Web Vitals thresholds have been measured to see around 24% lower abandonment during page loads.

You don’t need to memorise the studies. The mechanism is simple: every second of waiting filters out a slice of your visitors, and the visitors it filters out first are mobile users, in the Czech market, typically 60–70% of traffic to a local business site. A slow site is a shop where the door sticks. People don’t complain; they just don’t come in.

Why template sites are slow

If speed is worth this much, why is the average small-business website so slow? Because of how it was built. A typical WordPress site assembled with a page builder and a purchased theme ships:

• 2–4 MB of page weight: the theme’s every feature loads whether you use it or not
• 70–100+ requests: plugins, fonts, sliders, trackers, each fetched separately
• Render-blocking JavaScript: the browser must download and run code before it can show anything
• A database round-trip per visit: the page is assembled on the server, per visitor, on hosting shared with hundreds of other sites

None of this is visible in the sales demo, which you watched on fast Wi-Fi with a warm cache. It’s very visible at position 40 on PageSpeed Insights.

The alternative is old-fashioned and unglamorous: a hand-coded static site. The page is prepared once, at build time, and served as finished HTML: no database, no plugin stack, no assembly line between the visitor and the content. A static page doing the same job as that 3 MB template typically weighs under 200 kB and makes a dozen requests. It isn’t 10% faster; it’s a different category of thing. That’s the entire premise of how we build websites, and it’s why we can put speed guarantees in writing.

What a 100 PageSpeed score actually requires

Plenty of tools promise to “boost your score”: caching plugins, optimisation services, magic CDN layers. They can drag a slow site from 40 to 65. Genuine green-across-the-board scores aren’t an optimisation; they’re an architecture. The checklist:

1. Static HTML delivery: nothing computed per visit, ideally served from a CDN close to your visitors.
2. Almost no JavaScript: the biggest INP killer is script the page didn’t need. A brochure site needs kilobytes of it, not megabytes.
3. Images in modern formats (AVIF/WebP), sized to the layout, lazy-loaded below the fold, the most common LCP fix.
4. Fonts hosted on your own domain, preloaded, with fallbacks defined. This alone fixes most CLS problems.
5. Reserved space for every element: nothing may appear and push content around. Zero layout shift is a design decision, not a patch.
6. Third-party scripts on a strict diet: every chat widget and tracking pixel bills its cost in milliseconds. Each one must earn its place.

Two honest caveats, because scores get oversold. First, the number on PageSpeed Insights is a lab test; the field data above it is what Google actually ranks with, and it lags about a month behind any fix. Second, 100 is achievable and we do hit it, but the business goal is passing all three vitals in field data, not screenshotting a round number. A stable 98 in the field beats a lab-only 100 every day of the week.

Test your own site right now

Thirty seconds, no signup, no developer needed:

1. Open pagespeed.web.dev.
2. Enter your homepage address and run the test.
3. Look at the top section first: “Discover what your real users are experiencing”. That’s field data, the part Google uses. Check whether each vital says green.
4. Then check the Performance score below. Run it for Mobile, not desktop. Mobile is where sites fail and where most of your visitors are.
5. Repeat for your most important page (the service page or product page where customers actually decide) and, while you’re at it, for your two biggest competitors.

Reading the mobile score:

Score	What it means in practice
90–100	Genuinely fast, a competitive asset
50–89	The typical template site: measurably losing visitors
0–49	Speed is actively costing you customers today

If your site is too new or too small to have field data, the lab score and the vitals diagnostics below it are your guide. And if you have Google Search Console, the Core Web Vitals report there shows exactly which pages fail and why (free, straight from Google).

Bad scores: repair or rebuild?

What the test result usually means:

• Score 70–89, one vital failing: often fixable in place. Compressing images, removing unused plugins and deferring scripts are days of work, not weeks. Worth doing before anything drastic.
• Score below 50 on a template or page-builder site: the slowness is structural. You can spend CZK 20,000 optimising and plateau in the 60s, because the architecture is the bottleneck. At that point a rebuild on a fast foundation is usually cheaper than the sum of the patches, and it’s the only route that ends in green field data.
• Fast site, weak visibility: speed is necessary, not sufficient. If the vitals are green but customers still aren’t finding you, the gap is content and technical SEO, which is its own discipline.

Either way, decide from measurements, not from a sales pitch, including ours. Our pricing is public, so you can compare the cost of a rebuild against what the current site is quietly costing you each month.

Run the test, then tell us the number

Test your site at pagespeed.web.dev today: mobile tab, field data first. If the result is green, sincerely: nice work, you’re ahead of most of the Czech market. If it isn’t, send us your URL and we’ll reply within 24 hours with the three changes that would move your numbers most. Real diagnosis, no obligation attached.