Your website is slow on a phone, and that costs you calls

What does Google count as a slow website?

Google measures speed by what real visitors go through, in three numbers called Core Web Vitals (web.dev, Web Vitals, 31 Oct 2024; Google Search Central, Understanding Core Web Vitals, 10 Dec 2025).

• Largest Contentful Paint, or LCP, is how long the biggest image or block of text on screen takes to appear (web.dev, Largest Contentful Paint). Good is 2.5 seconds or less. Over 4 seconds is poor.
• Interaction to Next Paint, or INP, is how quickly the page reacts when someone taps a button or opens a menu (web.dev, Interaction to Next Paint). Good is 200 milliseconds or less. Over half a second is poor.
• Cumulative Layout Shift, or CLS, scores how much the page jumps while it loads (web.dev, Cumulative Layout Shift). Good is 0.1 or less. Over 0.25 is poor (Google for Developers, About PageSpeed Insights).

A page passes when 75 out of every 100 visits meet each target, with phones and computers counted separately (web.dev, Web Vitals).

Watch the phone result. Google uses the mobile version of a site’s content for indexing and ranking (Google Search Central, Mobile-first indexing best practices, 10 Dec 2025).

Google says Search “always seeks to show the most relevant content, even if the page experience is sub-par,” and that a great page experience helps where many pages answer a search well (Google Search Central, Understanding Google Page Experience, 22 Sep 2026). We think the bigger cost is the visitor who leaves.

Does a slow website really lose you customers?

Three companies measured it and published the results on web.dev, Google’s site for web developers.

• Vodafone sent half the traffic for one landing page to a faster version. A 31% better LCP brought 8% more sales and 15% more leads per visit (web.dev, Vodafone case study, 17 Mar 2021).
• Renault studied over 10 million visits in 33 countries. A page one second faster went with up to 14 fewer people in every 100 leaving after one page, most of all on pages already under 1.6 seconds (web.dev, Renault case study, 8 Oct 2021). That’s a pattern in the data, weaker evidence than a test.
• Nuvemshop, which runs more than 180,000 online stores in Latin America, changed how its stores load the first image on screen. A year later, mobile visitors from Google search ordered 8.9% more often (web.dev, Nuvemshop case study, 24 Jun 2026). Other changes in that year may have helped too.

These are large companies selling online. For a clinic or a contractor, the visitor who leaves is a call that never comes. We haven’t found published figures for smaller firms, so we treat these as a direction and measure each client’s own numbers.

How do you check your website’s speed in two minutes?

1. Go to pagespeed.web.dev, Google’s free PageSpeed Insights, and paste in your homepage address. Look at the mobile result.

2. The top section shows what real Chrome users got over the last 28 days: whether the page passes the Core Web Vitals assessment, and LCP, INP and CLS rated good, needs improvement or poor (Google for Developers, About PageSpeed Insights).

3. The score out of 100 below it comes from a lab test on a simulated mid-range phone on a mobile network. 90 or above is good, 50 to 89 needs improvement, under 50 is poor (Google for Developers, About PageSpeed Insights).

4. Repeat for the page most people reach from Google, usually your main service page.

What the results mean

• No real-visitor data means too few Chrome users visited. PageSpeed Insights then shows your whole site’s figures, or nothing (Google for Developers, About PageSpeed Insights), so go by the lab score.
• A poor LCP means the first screen is slow, usually because of the top image or the server.
• A poor CLS often comes from images with no set size or a banner that arrives late.
• A poor INP means the phone is too busy running scripts to answer a tap.

What makes small-business websites slow?

Four causes account for most of what we find.

Huge images

We often find a homepage photo of several megabytes, sized for a billboard and shown on a phone. web.dev’s fixes: compress images, use formats such as WebP and AVIF, serve smaller files to smaller screens, mark the top image with fetchpriority="high", and never lazy-load it, because that “will always lead to unnecessary resource load delay” (web.dev, Optimize Largest Contentful Paint, 31 Mar 2025). Give every image a width and height so the page holds its space (web.dev, Optimize Cumulative Layout Shift, 7 Feb 2025).

Page builders and heavy themes

In our experience, drag-and-drop builders and all-purpose themes load code for every feature they offer, used or not. web.dev traces slow taps to long tasks on the main thread, the single lane where the browser runs scripts and answers taps (web.dev, Optimize Interaction to Next Paint, 2 Sep 2025). Remove unused plugins and switch off features you don’t need. Some builders also write your text in with scripts, which hides it from AI assistants.

Too many scripts from other companies

Chat widgets, booking tools, tracking pixels, maps and videos each load code from another company’s server, and web.dev says they can keep the main thread busy. Its fixes: remove the ones that don’t earn their place, load the rest with async or defer, and load embeds such as videos only when someone reaches them (web.dev, Load third-party JavaScript, 19 Feb 2024).

Cheap or distant hosting

Nothing appears until your server sends its first byte. web.dev says hosting “should be the first thing you consider,” and that shared hosting is generally slower (web.dev, Optimize Time to First Byte, 28 Nov 2025). PageSpeed Insights rates a first byte slower than 1.8 seconds as poor (Google for Developers, About PageSpeed Insights). Distance adds time too; a content delivery network keeps copies of your site near your visitors (web.dev, Optimize Largest Contentful Paint). If your customers are in Dubai and Mumbai, ask your host where your server is.

What should you fix first?

We start with the top image on the homepage and main service pages, then the scripts, because both are cheap. Hosting comes next if the first byte is the slow part.

Judge the real-visitor numbers four weeks later, since they cover 28 days (Google for Developers, About PageSpeed Insights). The lab score shows the change straight away.