Core Web Vitals: why speed matters for your SEO

A fast site isn't just about pleasing search engines. Above all, it helps visitors understand your business, find the right information, and take action without waiting or dealing with a shifting page. Core Web Vitals provide three concrete benchmarks for observing this experience: main content loading, responsiveness to interactions, and visual stability.

Core Web Vitals, explained simply

Core Web Vitals are performance indicators focused on real user experience. They don't represent a site's entire quality and are no substitute for useful content, clean SEO structure, or a clear offer. Instead, they provide a common language to identify the main sources of friction.

LCP: does the main content display fast enough?

LCP stands for Largest Contentful Paint. It corresponds to the moment when the main content visible in the window — for example, a title, a large visual, or an introduction block — is effectively displayed.

To aim for a good experience, Google recommends an LCP of less than or equal to 2.5 seconds. Between 2.5 and 4 seconds, the experience needs improvement; beyond 4 seconds, it is considered poor. For a small business, the problem often stems from a main image that is too heavy, a slow server response, or resources loading before the useful content.

INP: does the site react when you click?

INP stands for Interaction to Next Paint. This indicator measures the delay between an interaction — touching a menu, opening a form, applying a filter, or clicking a button — and the moment the browser displays visible feedback.

A value less than or equal to 200 milliseconds corresponds to good responsiveness. Between 200 and 500 milliseconds, there is room for improvement; beyond 500 milliseconds, the interaction is likely to feel slow. A site can therefore display its page quickly but still be unpleasant to use if a large script monopolises the browser.

CLS: does the page stay in place?

CLS stands for Cumulative Layout Shift. It measures unexpected element shifts during loading. You're about to click a button, a banner appears, the content moves down: your click lands elsewhere. This is exactly the type of annoyance that this indicator helps to quantify.

A CLS of less than or equal to 0.1 is considered good. Between 0.1 and 0.25, stability can be improved; beyond 0.25, the shifts become too significant. Common causes include images without reserved dimensions, fonts that suddenly change the layout, and blocks that load late.

Thresholds to remember without needing a calculator

These thresholds are evaluated at the 75th percentile: we aren't just looking to measure the best visit, but to verify that the vast majority of users enjoy a proper experience. Results should also be read separately for mobile and desktop, as devices and networks do not offer the same conditions.

Speed, server, and Core Web Vitals: they aren't the same thing

A server response time is important, but it only tells part of the story. After the server responds, the browser still needs to download resources, build the page, display the content, and execute interactions.

This is also what distinguishes this article from a topic like Varnish. An HTTP cache can reduce the time needed to get a response and remains a useful lever. However, Core Web Vitals look at the complete experience: what the user sees, how quickly the page reacts, and the stability of its interface. Improving only the cache therefore does not guarantee a good LCP, INP, or CLS.

How to properly measure a site's performance?

Start with the pages that matter for your business: homepage, service page, portfolio or product, pricing, and contact. An isolated score on a secondary page doesn't tell you if the path leading to a quote request works well.

• PageSpeed Insights combines field data when enough real visits exist and a lab test performed with Lighthouse.
• Search Console groups URLs encountering Core Web Vitals issues and helps identify page types to address.
• Field tracking integrated into your statistics can supplement these tools if you have enough visits and if the data is correctly segmented.

Don't worry if two tools display different values. The lab simulates a device and a network under controlled conditions; field data reflects real devices, connections, and usage. The former helps diagnose and reproduce a problem. To prioritise what your visitors experience, give priority to field data when available.

Actions accessible to a small business

Improving LCP: display what matters first

Start by resizing images to the actual displayed size and by choosing a modern format like WebP or AVIF when context allows. Also, avoid loading several heavy visuals before the first useful screen. Preloading should be reserved for the truly priority resource, such as the main image.

The goal isn't to remove beautiful images or turn a page into a blank screen. It's about progressively delivering a readable page, then loading the rest without blocking the visitor. Response time and HTML generation count as much as file size.

Improving INP: lighten the browser's workload

Load non-essential scripts after the main content and prune tracking tools, widgets, and libraries that don't directly serve the user journey. Avoid very long processes at the moment of a click or a menu opening.

For a small business, this step often involves simplifying default-loaded features rather than rewriting the whole application. Test the result on a standard smartphone, not just on the creator's computer.

Improving CLS: reserve space before content arrives

Specify the dimensions or aspect ratio of images and videos. Also, plan space for banners, consent messages, and content that arrives late, so a delayed element doesn't push everything else down.

Finally, check fonts and loading strategies on mobile. A few pixels of movement can be enough to disrupt a click, even if the page seems stable on a large screen.

A simple five-step method

1. Choose three to five priority pages based on their commercial role, not based on how easy it is to get a good score.
2. Measure on mobile and desktop, distinguishing field data from lab data.
3. Identify the main bottleneck: main content too late, blocked interaction, or unstable layout.
4. Fix one family of causes at a time: images, scripts, fonts, components, or server time.
5. Measure again and verify that the improvement is reflected in real data, not just in a one-off test.

It's rarely useful to redo everything because a Lighthouse score hasn't reached 100. A page can get a good lab score and still be slow for some visitors; conversely, a detail flagged by the lab may have little commercial impact. The right goal is a fast, stable, and accessible experience on the journeys that matter.

Conclusion: performance is an experience, not a race for a score

Core Web Vitals give small and medium-sized businesses three easy-to-understand benchmarks: does the main content arrive fast enough, does the site react when prompted, and does the page stay in place? LCP, INP, and CLS thus transform a vague impression — "the site feels slow" — into concrete action plans.

Speed should be tracked as a matter of quality and conversion, alongside content, SEO, and accessibility. By starting with the pages that generate visits, calls, or quote requests, a small business can improve its site progressively, without sacrificing its identity and without adding unnecessary complexity.