February 8, 2026
Website speed is not just a technical detail. It directly affects how many visitors stay, how many leave, and how many become customers. Studies consistently show that every additional tenth of a second in loading time means lost revenue. And it is not just about e-commerce. Business websites, portfolios, and blogs. The same rule applies everywhere: a faster website converts better.
In this article, we will look at specific numbers, explain Core Web Vitals, walk through speed measurement tools, and provide a practical checklist you can use today.
According to a Google study, 53% of mobile visitors leave a website if it takes longer than 3 seconds to load. That is more than half of your potential audience, gone before they even see your content. Amazon found that every 100ms of delay costs 1% in revenue. At their scale, that means billions of dollars annually. And according to web.dev, Walmart recorded a 2% increase in conversions for every second they shaved off loading time.
For smaller businesses, the effect is even more pronounced. If your website takes 5 seconds to load, you are losing more than half of your visitors, and they are going to your competitors. If you run an online store with monthly revenue of $20,000 and speed up your site by one second, you can expect a revenue increase of $400 to $800 per month. That is money slipping through your fingers every single day.
Speed also affects search engine rankings. Google uses loading speed as one of its ranking factors. Slower websites appear lower in results, which means fewer organic visitors and higher advertising costs.
In 2021, Google introduced Core Web Vitals as key user experience metrics. Since then, they have become an important SEO factor. Let us break them down:
LCP (Largest Contentful Paint) measures how long it takes for the largest visible element on the page to appear, typically the main image or heading. Google considers values under 2.5 seconds as good. Between 2.5 and 4 seconds needs improvement. Above 4 seconds is poor. LCP is most affected by large images, slow-loading fonts, and render-blocking scripts in the page header.
INP (Interaction to Next Paint) replaced the older FID metric in March 2024. It measures how quickly the page responds to user interaction, including clicks, taps, and key presses. A good value is under 200 milliseconds. Between 200 and 500ms needs optimization. Above 500ms, the page feels uncomfortably slow. INP is primarily worsened by heavy JavaScript that blocks the browser main thread.
CLS (Cumulative Layout Shift) measures visual stability of the page. You know the experience: you are reading an article, and suddenly the text jumps because an ad banner loaded or an image without defined dimensions appeared. A good CLS value is under 0.1. Between 0.1 and 0.25 is problematic. Above 0.25 is poor. CLS is worsened by images without width and height attributes, dynamically injected elements, and web fonts that shift layout when they load.
You can track all three metrics in Google Search Console under the Core Web Vitals section. Google evaluates pages based on data from real users (called field data), so the results reflect the actual visitor experience.
Before you start optimizing, you need to know where you stand. Several tools can help:
Google PageSpeed Insights is the most important tool. It shows you a score from 0 to 100 and specific recommendations for improvement. It displays both lab data (simulated test) and field data from real users. Focus primarily on field data, as that is what determines SEO impact.
GTmetrix offers more detailed analysis with a waterfall chart of loading. You will see exactly which files load in what order and how long they take. It is great for diagnostics. When PageSpeed says you have an LCP problem, GTmetrix shows you exactly why.
WebPageTest is an advanced tool that lets you test from different locations and on different devices. You can simulate slow connections, compare multiple pages side by side, and analyze the loading filmstrip frame by frame. It is free and incredibly detailed.
Lighthouse in Chrome DevTools (press F12, Lighthouse tab) runs directly in your browser. It uses the same engine as PageSpeed Insights but runs locally. It is useful for quick testing during development.
I recommend measuring regularly, ideally after every major change to the website. Save your results so you can track the trend over time.
Images make up over 50% of the total page weight on the average website. Optimizing them is therefore the most effective step you can take.
Next-generation formats: WebP offers 25-35% smaller files than JPEG at the same quality. AVIF goes even further with 50% savings compared to JPEG. Both formats are supported by all modern browsers. If you need to support older browsers, use the <picture> element with a JPEG fallback.
Responsive images: There is no point sending a 1920px-wide image to a phone with a 375px screen. The srcset attribute and <picture> element let you serve different sizes for different devices. The browser automatically chooses the smallest sufficient variant.
Lazy loading: Images below the fold do not need to be downloaded immediately. The loading="lazy" attribute tells the browser to download the image only when the user is scrolling near it. This significantly speeds up initial page load. Be careful, though: the first visible image (LCP candidate) should not have lazy loading, as this would paradoxically worsen LCP.
Always specify dimensions: Every <img> tag should have width and height attributes. This lets the browser reserve space in the layout before downloading the image, preventing content shifts (CLS).
Tools like Squoosh or Sharp can help you easily convert and compress images.
JavaScript is ubiquitous on modern websites, but every kilobyte of JS costs more than a kilobyte of images. The browser must not only download JS but also parse, compile, and execute it, all of which blocks the main thread.
The defer and async attributes: Scripts in the page header without these attributes block rendering. The defer attribute tells the browser to download the script in the background and execute it only after the HTML is fully processed. The async attribute downloads the script in the background and executes it immediately after download. It is suitable for independent scripts like analytics.
Remove unused code: The Chrome DevTools Coverage tab shows what percentage of downloaded JS is actually used. On many websites, it is only 30-40%. The rest consists of unused functions, polyfills for old browsers, and dead code from removed features.
Third parties: Chat widgets, tracking scripts, A/B testing tools, and social plugins. Each adds tens to hundreds of kilobytes of JS. Consider whether you truly need each one. Often you will find that a chat widget nobody uses is slowing your site by 800ms.
Minification and compression: Minification removes unnecessary whitespace, comments, and shortens variable names. Gzip or Brotli compression then further reduces the transferred file by 60-80%. Both should be handled by your build process or server automatically.
Hosting is the foundation upon which your website speed rests. You can have perfectly optimized code, but if the server takes 2 seconds to respond, nothing else matters.
Shared hosting is the cheapest option, but your website shares resources with thousands of others. When a neighbor on the server runs a flash sale and their site consumes all the CPU, your site suffers too. For business websites, I recommend at least a VPS (Virtual Private Server), where you have guaranteed resources.
TTFB (Time to First Byte) is a metric that measures how quickly the server starts responding. A good value is under 200ms. If your TTFB is above 600ms, hosting is holding you back. Measure it in DevTools under the Network tab. Look at Waiting (TTFB) for the first HTML document.
CDN (Content Delivery Network) distributes copies of your website to servers around the world. A visitor from Prague gets data from the nearest server instead of from a server in America. CDNs like Cloudflare offer a free plan that speeds up your website for international visitors and adds DDoS protection. For local businesses with local clientele, a CDN is less critical but still brings benefits through caching and compression.
HTTP/2 and HTTP/3: Modern protocols allow downloading multiple files simultaneously over a single connection. Most hosting providers today support HTTP/2, but verify it. HTTP/3 with the QUIC protocol is even faster, especially on mobile networks with higher latency.
A practical list you can go through step by step:
loading="lazy" to images below the fold<img> has width and height attributesCache-Control and Expires headersI recently optimized a website for a client whose pages had a PageSpeed score of 28 on mobile. Loading took over 8 seconds. After analysis, I identified three main problems: unoptimized PNG images (12 MB total on the homepage), 14 unused JavaScript libraries, and shared hosting with TTFB over 1.5 seconds.
I converted images to WebP, added lazy loading, removed unnecessary scripts, and moved the website to faster hosting. The result: a score of 94, loading under 1.5 seconds, LCP of 1.2 seconds. Over the following three months, the client saw a 34% increase in organic traffic and a 22% increase in contact form conversions.
The key point is that speed optimization is not a one-time task. Websites gradually slow down as you add new content, images, scripts. That is why I recommend regular checks at least once a month.
All websites I build are optimized for speed from the ground up, with no unnecessary dependencies, compressed files, proper caching. If you want to know how your website performs, get in touch. I will evaluate it for free.
Related articles: When It’s Time for a New Website • WordPress vs. Custom Website