PageSpeed Insights: understand and improve your score in 2026
The complete guide — not just the score, but Core Web Vitals, field data and the real optimizations that matter.
- PageSpeed Insights doesn't just measure "speed" — it runs a full Lighthouse audit covering performance, accessibility, best practices and SEO
- Core Web Vitals (LCP, INP, CLS) are the metrics that actually matter for Google — and they're measured on real users, not in a lab
- A 100/100 score is not the goal — a site that's fast for your users is. Orilyt uses PSI data in 3 critical tests to give you an actionable view
When it comes to web performance, PageSpeed Insights (PSI) is often the first tool mentioned. A red score causes panic, a green score brings relief. But behind that number, what does PSI actually measure? And more importantly, what should you do about it?
The reality is that most WordPress site owners look at the score without understanding what it represents. They optimize for the number rather than for user experience. And they miss what actually matters: Core Web Vitals.
This guide goes beyond the score. We'll break down what PageSpeed Insights really measures, explain each Core Web Vital in plain language, and give you the 5 optimizations with the most impact — not the 50 that nobody applies.
What PageSpeed Insights actually measures
PageSpeed Insights is not a simple speed test. It's an interface that combines two completely different data sources: lab data (Lighthouse) and field data (Chrome UX Report).
Lighthouse is an automated audit that runs in a controlled environment. It simulates a mid-range mobile device on a slow 4G connection. That's where the famous 0-to-100 score comes from.
The Chrome UX Report (CrUX), on the other hand, collects anonymized data from real Chrome users over the past 28 days. It's reality — not a simulation.
Lighthouse audits 4 categories: performance (the main score), accessibility, best practices and technical SEO. Each category has its own score, but it's the performance score that gets all the attention.
Core Web Vitals: the 3 metrics that matter for Google
Since 2021, Google has used Core Web Vitals as a ranking signal. These are not lab metrics — they're measurements from real users. Here are all three, in plain language.
LCP measures the time it takes to display the largest visible element on the page (often a hero image or heading). It's the user's perception of "the page has loaded."
INP replaced FID (First Input Delay) in March 2024. It measures page responsiveness: when the user clicks, types or scrolls, how long does it take for the page to visually react?
CLS measures visual stability. You know that frustrating moment when you're about to click a button and the content shifts because an ad or image loads? That's exactly what this measures.
Lab vs field data: why your scores diverge
This is the most common confusion with PSI. Your Lighthouse score says 45, but your Core Web Vitals are all green. How is that possible?
Lab data (Lighthouse) simulates a worst-case scenario: a mid-range smartphone on a slow 4G connection, no cache. It's useful for identifying technical issues, but it doesn't represent your actual users.
Field data (CrUX) reflects the real experience of Chrome visitors on your site. If most of your users are on desktop with fiber, field data will be much better than the lab.
The key: Google uses field data (Core Web Vitals) as a ranking signal, not the Lighthouse score. The lab score is still useful for diagnostics, but it's not what influences your SEO.
The 0 to 100 score: what each zone means
The Lighthouse performance score is calculated from 5 weighted metrics. Here's what each zone means in practice.
Note: the score can vary by 5 to 10 points between consecutive measurements, even with no changes. This is normal — network and server conditions fluctuate. Don't panic over a single point drop.
How Orilyt uses PageSpeed data
Orilyt integrates PageSpeed Insights data into 3 critical audit tests. Rather than leaving you alone with a 50-line Lighthouse report, Orilyt extracts what matters and turns it into actionable recommendations.
- Test #01 — TTFB (Time to First Byte): measures server response time. If your host takes over 800 ms to respond, no front-end optimization will compensate.
- Test #02 — Page Weight: analyzes the total weight of resources (HTML, CSS, JS, images, fonts). An average WordPress site weighs 3 MB — the goal is to stay under 2 MB.
- Test #03 — Mobile Friendly: verifies that the site displays correctly and is usable on mobile. 60% of web traffic is mobile — this is no longer optional.
Each test produces an FIA recommendation (Fact, Impact, Action): a factual finding, its business impact, and the concrete corrective action. This is what turns a score into a selling point for your clients.
5 quick optimizations with the most impact
Among the dozens of recommendations PSI can generate, these 5 actions cover 80% of the performance gains on a typical WordPress site.
1. Optimize images
Convert to WebP, resize correctly and enable lazy loading. This is almost always the biggest gain — often 30 to 50% of page weight.
2. Enable compression (Gzip/Brotli)
Compression reduces the size of text files (HTML, CSS, JS) by 60 to 80%. It's a server configuration, often a single line in .htaccess.
3. Leverage browser caching
Set cache headers (Cache-Control, Expires) so static resources aren't re-downloaded on every visit. Immediate impact on returning visitors.
4. Defer non-critical JavaScript
Scripts that block rendering (analytics, widgets, sliders) should be loaded with defer or async. The browser shows content first, then runs the scripts.
5. Reduce server response time
A TTFB above 800 ms signals a server-side problem: overloaded shared hosting, slow SQL queries, or no object cache. Switching hosts can change everything.
3 PageSpeed Insights myths to forget
Myth #1: "A 100/100 score is the goal"
No. The goal is a fast site for your users, not a perfect lab score. A complex e-commerce site at 72 can offer a better experience than a brochure site at 98 if its Core Web Vitals are green.
Myth #2: "The PSI score is a Google ranking factor"
The Lighthouse score is not an SEO signal. It's the Core Web Vitals (field data) that count, and they're just one signal among hundreds. Quality content with a score of 60 will always beat mediocre content at 95.
Myth #3: "Mobile and desktop scores should match"
The mobile test simulates a mid-range smartphone on a slow connection. A 30-40 point gap is normal. If your mobile score is above 60 and Core Web Vitals pass, you're in good shape.
What changed in 2025-2026
The PageSpeed landscape is constantly evolving. Here are the most important recent changes:
- INP replaces FID: since March 2024, Interaction to Next Paint is the new responsiveness Core Web Vital. It measures all interactions, not just the first one.
- Lighthouse 12: new diagnostic metrics, better detection of layout shift issues related to fonts and CSS animations.
- Adjusted weighted scoring: TBT (Total Blocking Time) weight has been increased, reflecting the growing importance of responsiveness.
- PSI and AI: Google is starting to integrate context-aware suggestions based on machine learning into PageSpeed recommendations.
Beyond the score: user experience
PageSpeed Insights is a powerful tool, but you need to know how to read it. The lab score is a diagnosis, not a verdict. Core Web Vitals are the real signal. And optimizations should target user experience, not the number.
For freelancers and agencies, understanding PSI in depth is a competitive advantage. When a client says "my score is red," you can explain why, what really matters, and propose a concrete action plan.
Orilyt automates this analysis: TTFB, page weight and mobile friendly tests built into every audit, with FIA recommendations ready to present. No need to decipher Lighthouse yourself.