When we show a Lighthouse report to a Geneva business owner for the first time, the question is nearly always the same: "What does 42 mean?" The short answer: your site is losing Google rankings, visitors and customers every day.
Lighthouse is Google's official audit tool. A score of 95 or more on Performance is not a badge for developers. It is the threshold at which a website stops holding back its own visibility.
This article explains what Lighthouse is, how to read its score, why 95+ is the minimum for a professional website in 2026, what drags scores down (WordPress, plugins, images, unnecessary JavaScript) and what pushes them up (custom-built code, clean architecture).
Lighthouse: what it is and why Google built it
The web's reference audit tool
Lighthouse is an open-source tool developed by Google and built into Chrome DevTools since 2017. It analyses any public URL across five areas:
Each area is scored from 0 to 100. The Performance score is the one that directly affects your Google rankings, and it is the subject of this article.
How Lighthouse calculates its score
The Performance score is a weighted combination of five lab metrics:
INP (Interaction to Next Paint), which replaced FID in the Core Web Vitals in 2024, is not part of the Lighthouse score: it is measured on real visits, in Search Console. TBT is its closest lab equivalent. A low TBT usually points to a good INP.
Lighthouse or field data (CrUX)
A point that is often misunderstood: Lighthouse measures simulated conditions (slow 4G, CPU slowed down 4×). That is not what your real visitors experience.
Field data (the Chrome User Experience Report, CrUX) measures real Chrome users. Google Search Console shows it in the "Core Web Vitals" section. A site can score 88 on Lighthouse and still be rated "Good" in the field, or the other way round.
In practice: with a Lighthouse score of 95+, your field data is very likely to sit in the "Good" zone too. Below that, it gets blurry: some visitors have a good experience, others do not, and Google takes that into account.
Why 95+ is the minimum in 2026
Expectations have changed, and so has the algorithm
In 2026, 95+ has become the credibility threshold for a professional website, for three reasons.
Google Page Experience: since its full rollout in 2022, page experience has been a confirmed ranking factor. Sites that miss the "Good" thresholds on Core Web Vitals start at a disadvantage, whatever the quality of their content.
The cumulative effect: Google does not apply a one-off penalty, it keeps favouring fast sites. Month after month, a fast site builds a lead over a slow one in impressions and clicks.
Visitors are less patient: on mobile, a few seconds of waiting are enough for a visitor to go back to the next result. Swiss users, used to fast connections, have little patience for slow sites.
Why aim for 95, not 100
A score of 100 is possible but rare in production (with analytics scripts, third-party fonts and chat widgets). You get it on very stripped-down static pages or under perfect lab conditions.
95+ is realistic, sustainable and enough to secure every documented SEO benefit. Below 90, you lose those guarantees. Between 90 and 94, you are in a grey zone. From 95 up, you are in the clear.
What drags scores down: the usual suspects
WordPress and its plugins: built-in technical debt
WordPress powers 43% of the world's websites (W3Techs, January 2025). It made website creation accessible, but it penalises performance.
Premium themes: Divi, Elementor, Avada, WPBakery. These page builders load a large volume of CSS and JavaScript, of which only a small part is used on any given page. The browser still has to parse, compile and evaluate all of it before displaying anything.
A Lighthouse audit of a WordPress site built with Elementor almost always flags the same issues: render-blocking resources, unused JavaScript and CSS, JPEG images instead of WebP or AVIF, poorly compressed images. Not because WordPress is broken, but because the stacked layers (theme, page builder, plugins) add weight that cannot be removed without rewriting the site.
Security, SEO and analytics plugins: the quiet overload
Every WordPress plugin adds code. Server-side PHP, but often also JavaScript and CSS that run in every visitor's browser.
A "standard" WordPress site for a Geneva SME quickly piles up active plugins. You often find:
- 1 SEO plugin (Yoast, Rank Math): adds JS for real-time validation
- 1 security plugin (Wordfence): adds client-side detection JS
- 1 analytics plugin (MonsterInsights): loads a Google Analytics script plus a wrapper
- 1 caching plugin (WP Rocket, W3 Total Cache): helps, without offsetting the others
- 1 form plugin (Contact Form 7, Gravity Forms): loads its JS on every page
- 1 GDPR/cookie plugin: heavy JavaScript overlay, often render-blocking
On their own, these scripts are not a problem. Together, they make the page much heavier and multiply HTTP requests on load. A caching plugin does not fix that; it softens the effect a little.
Unoptimised images: the main cause of a poor LCP
According to the HTTP Archive report (2024), images make up on average 51% of a web page's weight. On WordPress sites, their share is often heavier still, because:
- Images are uploaded as full-resolution JPEG or PNG files, straight from the camera
- They are not resized (a phone receives the same image as a 4K screen)
- They are not converted to WebP or AVIF, formats that are much lighter than JPEG
- Lazy loading is not always set up correctly
- The hero image (the one that determines LCP) is not preloaded
A single change, switching the hero image from a 2 MB JPEG to a 300 KB AVIF, can be enough to bring LCP into the green and lift the Lighthouse score clearly.
Third-party JavaScript: a gradual slowdown
Third-party scripts are among the main enemies of performance. Chat widget, Facebook pixel, A/B testing script, Google Maps in an iframe: each one adds requests to external servers whose response times you do not control.
They also create Long Tasks on the browser's main thread. Those are the tasks that degrade INP.
What pushes scores up: clean architecture
Next.js 16: a foundation designed for performance
Next.js is not magic. But its default architecture solves at the source the problems that drag Lighthouse scores down:
Hybrid rendering: a page can be generated as static HTML (SSG), rendered on the server for each request (SSR), or a mix of both. Either way, the browser receives HTML that is ready to display, not an empty document waiting for JavaScript to run.
Server Components (React 19): purely presentational components (text blocks, static images, navigation) generate no client-side JavaScript. The bundle gets much lighter and the hydration tasks that weigh on INP disappear.
Turbopack: the native bundler in Next.js 16 produces optimised bundles with automatic code splitting. Each page loads only the JavaScript it needs.
next/font: Google Fonts are downloaded at build time and served from your own domain. No third-party requests on load, no font-related CLS, no extra network delay.
Tailwind v4: only the CSS you use
Tailwind CSS v4 generates a CSS file containing only the classes present in the source code. Production CSS therefore stays very light, far below a Bootstrap or WordPress theme loaded in full.
That matters: lighter CSS means less parsing, fewer reflows and a faster FCP. Combined with @layer to prioritise critical styles, Tailwind v4 removes the "Reduce unused CSS" audit from Lighthouse altogether.
GSAP: animations with no performance cost
GSAP (GreenSock Animation Platform) is the standard animation library on INFUSE projects. Its reputation for being lightweight is deserved:
- GSAP uses
requestAnimationFrameto sync animations with the browser's rendering cycle, without layout thrashing. - GSAP animations target
transformandopacityfirst, GPU-accelerated properties that do not trigger reflow and do not affect CLS. - GSAP is imported with tree-shaking: only the modules in use (ScrollTrigger, SplitText, etc.) end up in the bundle.
Result: smooth 60 fps animations with no effect on LCP, INP or CLS.
Hosting and infrastructure: the often forgotten factor
The best code, served by a slow host, is still slow. A TTFB (Time to First Byte) above 600 ms already uses 25% of the LCP budget before the browser shows a single pixel.
INFUSE projects are deployed on Vercel, whose global CDN serves static files (images, fonts, CSS) from the node closest to the visitor. TTFB therefore stays low from Switzerland.
Practical audit: reading your Lighthouse score today
In 5 minutes, with nothing to install
- Open Chrome > F12 (or Cmd+Option+I on Mac) > "Lighthouse" tab
- Tick "Performance", "SEO", "Best practices"
- Select "Mobile" (Google's reference environment)
- Click "Analyse page load"
Or go to pagespeed.web.dev: enter your URL and get results in 30 seconds.
How to interpret the results
The diagnostics that tell you most
Lighthouse gives you more than a score: it also gives you an action plan. The three diagnostics to check first:
"Eliminate render-blocking resources": scripts and CSS loaded in blocking mode before the page renders. A sign of a poor loading order, and often the biggest loss of points.
"Reduce unused JavaScript": the ratio of JS loaded to JS executed. The larger the unused share, the bigger the architecture problem. On WordPress themes, it is often very high.
"Serve images in next-gen formats": if your images are JPEG or PNG, switching to WebP or AVIF cuts their weight clearly with no visible loss. Next.js does this automatically.
Lighthouse score and SEO: a documented link
What Google has confirmed
Google announced in 2021 that page experience, including the Core Web Vitals, is a ranking factor. Google says clearly that pages offering a good experience are more likely to rank well.
Less documented, but seen in practice: the effect builds up over time. A site that keeps excellent metrics over the long run sees its authority grow faster, which strengthens every new piece of content it publishes.
The effect on Google Ads Quality Score
If you advertise on Google, the Quality Score of your landing pages depends partly on landing page experience. A fast page improves that part of your Quality Score.
A better Quality Score lowers your cost per click. On a regular advertising budget, that drop can cover part of the cost of a custom-built site.
The effect on crawl budget
Google's bots visit your site continuously to index new pages. A slow site needs more server resources for each crawler visit, and fewer pages get indexed per session.
A high Lighthouse score goes hand in hand with a low TTFB and light pages. Fast sites are crawled more efficiently and new content is indexed sooner.
Keeping a 95+ score over time
Monitor continuously
A Lighthouse score is never permanent. Add a new script, a third-party widget or an oversized image, and the metrics drop.
Good habits:
- Audit on every deployment: add Lighthouse to your CI/CD pipeline with Lighthouse CI (open source). An alert fires automatically if a metric drops.
- Watch third-party scripts: assess the impact of every new script (analytics, chat, A/B testing). Systematic lazy loading, deferred loading after the first render.
- Check images on upload: set an internal rule: no image above 200 KB before optimisation, WebP or AVIF only.
Additions that cost nothing
Some practices improve the experience without affecting the metrics:
- Preconnect to critical third-party domains (fonts, analytics) via
<link rel="preconnect"> - Prefetch pages likely to be visited next (Next.js does this automatically via
<Link>) - Service Worker to cache resources between visits (can be configured with Next.js in PWA mode)
The Lighthouse score as a quality standard
Aiming for 95+ on Lighthouse is not a technical fad. It is the measurable sign of a fast, stable and responsive site. Google measures it, your visitors feel it, and conversion rates show it.
The gap between a site at 45 and one at 95 shows up in results. In Geneva and across French-speaking Switzerland, where companies are starting to take organic search seriously, it is a concrete chance to get ahead of competitors who have not yet made the move.
At INFUSE, 95+ is the starting point, not the target. Every site passes the audit before going live. Below 95, it does not ship.
Frequently asked questions
What Lighthouse score should a professional website aim for?
95 or more on Performance. A score of 100 is rare in production with analytics scripts or third-party fonts, and 95+ is enough to secure the documented SEO benefits.
Does a good Lighthouse score guarantee good field data?
Lighthouse measures simulated conditions, while CrUX data comes from real Chrome users. At 95+, field data is very likely to sit in the "Good" zone too. Below that, the two can diverge.
How can I test my site's Lighthouse score?
In Chrome, open DevTools (F12), go to the "Lighthouse" tab, select "Mobile" and run the analysis. You can also enter your URL on pagespeed.web.dev and get the result in 30 seconds.
Want to know where your site stands? Request a free audit. Within 48 hours, you receive a full report: current metrics, issues identified and a costed action plan.