How fast must a website be on a phone? A 2-minute check

Picture a customer with a burst pipe. They search on their phone, tap your site and wait. If the phone number takes ten seconds to appear, they go back and tap the next result. Speed is not a technical detail here. It decides whether the phone rings.

The three numbers, in plain words

Google publishes three measures it calls Core Web Vitals. According to web.dev, Web Vitals (updated 31 October 2024), a page counts as good when 75 % of visits reach these values, with mobile and desktop judged separately.

The limits come from the LCP (updated 4 September 2025), INP (updated 2 September 2025) and CLS pages. For a small firm, LCP is the one that matters most. It is the moment the customer sees your headline or photo instead of a blank screen. CLS is the annoying one: you go to tap "Call" and the button slides away because an image loaded above it.

What slows down small-firm sites

web.dev's guide to optimising LCP lists the usual causes for the main image: the file is large or in an old format, the image is lazy-loaded although it is the first thing on screen, and style sheets or scripts block the page from drawing. From that and from the CLS page, these are the patterns to look for:

• Huge photos. A photo straight from a phone camera is usually much bigger than any screen needs. Resize it to the displayed size and use a modern format such as WebP or AVIF.
• Sliders. A slider is several large photos in one place, and the first one is usually the biggest thing on the screen, so it sets your LCP.
• Chat widgets and pop-ups. They add a script from another company. Google's CLS page names third-party widgets that resize themselves as a cause of jumping.
• Too many fonts. Each font is another file to download, and a font that renders larger than its fallback shifts the text when it arrives (also from the CLS page).
• Images without set dimensions. The browser does not know how much room to leave, so the text below jumps.
• A stack of scripts. Analytics, pixels, booking tools and plug-ins, each one small, together heavy.

The 2-minute check

1. Open pagespeed.web.dev on a computer.
2. Paste your web address and press Analyze. It takes up to a minute.
3. Open the Mobile result. A customer standing in a queue or a car park is on a phone, not a desktop.
4. Look at the Core Web Vitals box first. Google explains that it shows data from real visitors over the last 28 days. A small site may not have enough visits, and then the box is empty.
5. Find three lines: Largest Contentful Paint, Cumulative Layout Shift and, in the real-visitor box, Interaction to Next Paint. Compare them with the table above. Green is good.
6. Read the list of suggestions under the score. The first few usually name the heavy images or scripts.
7. Now the human test: switch Wi-Fi off, open your site on mobile data and count seconds until you can read your phone number. This is a stopwatch, not an exact LCP, but if you count past five, you have your answer.

One caution from the same Google page: the lab score comes from one simulated phone and connection, so it is a guide, not a verdict. Test the home page and your most visited service page.

How we measured ours

In October 2026 we tested the websites of 53 heating firms in Bratislava with Lighthouse 13.5, the engine behind PageSpeed Insights, using a simulated mid-range phone. The median site showed its main content after 6.3 s. Our sample site Teplo Váh, an invented firm, showed it after 2.1 s (measured 31 October 2026).

This is our own measurement and it has limits. It is a lab simulation on one computer, not real visitors. The 53 sites are one trade in one city. Teplo Váh is a sample we built, not a customer result. The method and numbers are on our Optimisations page.

Does speed pay? A study by 55 and Deloitte for Google of mobile load times over 30 days at the end of 2019 found that a 0.1 s faster mobile site raised conversion in retail by 8.4 %. That is online shops, not plumbers, and the page has no date, so read it as a direction, not a promise.

How to read the result

A simple decision rule. If LCP is green and CLS is green, leave it alone and spend your time elsewhere. If LCP is amber or red, the cause is nearly always the first photo or a script, so ask for those two to be fixed first. If only CLS is red, ask for image sizes to be set and for the chat widget to be checked. Fix one thing, run the check again and compare.

What to ask your web person

• What is our mobile LCP, CLS and INP today, and where did you measure it?
• What is the largest image on the home page, in kilobytes?
• Is the first photo lazy-loaded? It should not be.
• Which outside scripts and widgets load, and can each one go?
• How many fonts do we load?
• Can you send a before and after for the two pages that matter most?

If you would rather have speed handled from the start, it is included in all our packages, listed on the prices page. A site's speed is also one of the things worth checking next to your Google Business Profile.