Google PageSpeed Insights analyzes any URL for free and gives you real-world performance data along with a prioritized list of improvements. To use it, go to pagespeed.web.dev, enter your page URL, and click Analyze. Within seconds you'll get both field data (real users) and lab data (controlled simulation), plus concrete recommendations.
This guide explains what each section means, how to read the results, and which actions to take to effectively raise your score.
How PageSpeed Insights Works: Field Data vs. Lab Data
Before looking at the numbers, understand where they come from:
- Field data (CrUX): real experiences from Chrome users on your page over the past 28 days. This is the data Google uses for ranking. It only appears if your page has enough traffic.
- Lab data (Lighthouse): a simulation run on a throttled network (slow mobile by default). It always appears, but doesn't represent real conditions for all your users.
The 0–100 score you see on screen comes from lab data. A site can score 55/100 in the lab and still have excellent Core Web Vitals in the field if the server and CDN are fast. Look at both sections, not just the number.
The Key Metrics You Need to Understand
PageSpeed Insights reports several metrics. The most important are the Core Web Vitals:
| Metric | What It Measures | "Good" Target |
|---|---|---|
| LCP (Largest Contentful Paint) | How long it takes to render the largest visible element (hero image, main block, title) | ≤ 2.5 s |
| INP (Interaction to Next Paint) | Response latency to user interactions (clicks, taps, keystrokes) | ≤ 200 ms |
| CLS (Cumulative Layout Shift) | Visual stability — how much elements "jump around" as the page loads | ≤ 0.1 |
| FCP (First Contentful Paint) | Time until the first text or image appears on screen | ≤ 1.8 s |
| TBT (Total Blocking Time) | Time the main thread is blocked and unable to respond to the user | ≤ 200 ms |
| Speed Index | How quickly the page fills visually | ≤ 3.4 s |
The three Core Web Vitals (LCP, INP, CLS) are direct ranking signals. Prioritize them above all other metrics.
How to Read Opportunities and Diagnostics
Below the metrics you'll find two sections:
Opportunities
These are improvements with an estimated load-time saving. PageSpeed ranks them by impact. The most common ones are:
- Eliminate render-blocking resources — external CSS and JS in the
<head>that delay the FCP. - Serve images in next-gen formats — converting JPEG/PNG to WebP or AVIF.
- Efficiently encode images — uncompressed images that weigh far more than needed.
- Defer offscreen images — images below the fold loading at the same time as above-the-fold content.
- Enable text compression — GZIP or Brotli not yet enabled (check our web performance guide for the exact steps).
- Reduce server response time (TTFB) — the server takes too long to send the first byte.
Diagnostics
These are performance issues without an exact estimated saving: inefficient cache policies, unused JavaScript, unused CSS, excessive DOM size, and more. They're still worth addressing systematically.
The Right Workflow for Improving Your Score
Don't try to fix everything at once. Follow this order:
- Analyze mobile first. Google indexes in mobile mode; the desktop score is always better and represents a lower risk area.
- Copy the opportunities with the highest estimated savings to a spreadsheet or task list.
- Apply one improvement at a time — this lets you measure the real impact of each change.
- Re-analyze after each change to confirm improvement and catch any regressions.
- Repeat until you hit 90+ on mobile, or until the remaining estimated savings are under 0.1 s.
Specialized agencies like elenlace.com follow exactly this iterative process before declaring a site fully optimized.
Concrete Fixes Based on What You Find
High LCP (image or block takes too long to appear)
- Add
rel="preload"to your hero image:<link rel="preload" as="image" href="hero.webp">. - Convert the image to WebP and set its dimensions to match the actual display size.
- Never apply
loading="lazy"to the image that's visible on initial load.
High CLS (elements shifting on load)
- Always define
widthandheightattributes on your<img>tags. - Reserve space for ads and embeds with a fixed-size container.
- Load web fonts with
font-display: swapto prevent text jumps.
High TBT / INP (page feels sluggish or unresponsive)
- Identify your heaviest third-party scripts and load them with
deferorasync. - Break long JavaScript tasks into smaller chunks.
- Use Web Workers for intensive computation off the main thread.
Render-blocking resources
- Move non-critical
<script>tags to the end of the<body>, or adddefer. - Extract critical "above the fold" CSS and inline it in the
<head>; load the rest asynchronously.
Key Takeaways
- PageSpeed Insights combines real-user data (CrUX) with lab simulations (Lighthouse) — look at both sections.
- Core Web Vitals (LCP, INP, CLS) are direct ranking signals — prioritize them above all else.
- Always analyze in mobile mode first.
- Apply improvements one at a time and re-measure to quantify the real impact.
- The fastest wins are usually: compress images, enable GZIP, defer non-critical JS, and add browser caching.
Improving your PageSpeed Insights score is an iterative process, not a one-time fix. If you need professional help diagnosing and resolving your site's performance bottlenecks, the team at elenlace.com can run a complete audit and apply the fixes for you. Explore more optimization techniques in our web performance section.
FAQ
Do I need to reach 100/100 on PageSpeed Insights?
It's neither necessary nor realistic for most sites with dynamic content, third-party plugins, or advertising. A score of 90+ on mobile is a solid goal. What matters most is that your Core Web Vitals are in the "Good" range according to field data, since those are the signals Google actually uses for ranking.
Why does my score change between consecutive analyses?
Lighthouse lab data varies because it simulates network and CPU conditions that can fluctuate slightly between runs. For a reliable trend, average 3–5 consecutive analyses or track your score regularly over time using the PageSpeed Insights API or a monitoring tool.
Does PageSpeed Insights measure mobile or desktop?
Both, separately. The default tab when results load is Mobile, which simulates a slow 4G connection and a mid-range CPU. You can switch to Desktop with the selector at the top. Always prioritize the mobile results.
How long does it take for an improvement to show up in field data?
Field data (CrUX) updates on a rolling 28-day window. This means you'll see the full impact of your changes in the report about 4 weeks after implementing them. Lab data reflects changes immediately on the next analysis run.
Further reading
Other providers and guides worth comparing:
