How to Achieve 100/100 on PageSpeed Insights in 2026

How is the PageSpeed Insights performance score calculated?

PageSpeed Insights runs Lighthouse against your page and turns five lab metrics into a single score from 0 to 100. Each metric is scored against real-world data from the HTTP Archive and then weighted. Because three metrics carry 80% of the weight, they are where your effort pays off.

Scores of 90–100 are rated good, 50–89 need improvement, and 0–49 are poor. The mobile test emulates a mid-range phone with a slowed-down CPU on a throttled 4G connection, which is why a site that feels instant on a laptop can still score in the 60s on mobile.

Step 1: Make the server respond in under 200 ms

Every other metric waits for the first byte of HTML. If Time to First Byte (TTFB) is slow, FCP, LCP and Speed Index are all slow too.

• Statically generate or cache full HTML responses at a CDN edge wherever the content allows.
• Keep database queries off the critical path of public pages: render from a cache and revalidate in the background.
• Enable HTTP/2 or HTTP/3 and Brotli compression on your host or CDN.
• Avoid redirect chains — each redirect adds a full network round trip before the page can start loading.

Step 2: Fix Largest Contentful Paint (LCP)

The LCP element is usually a hero image or a large heading. The Lighthouse report names it under "Largest Contentful Paint element". Your goal is to have the browser discover and download it as early as possible.

• Serve hero images in AVIF or WebP, sized to the dimensions they are actually displayed at.
• Mark the LCP image as high priority with fetchpriority="high" (or the priority prop on the Next.js Image component).
• Never lazy-load the LCP image, and don't hide it behind a client-side fade-in animation.
• Preload the font used by above-the-fold text and use font-display: swap or optional so text paints immediately.

Step 3: Reduce Total Blocking Time (TBT)

TBT adds up the time beyond 50 ms of every long task on the main thread between first paint and interactivity. It is the heaviest-weighted metric and is almost always caused by JavaScript.

• Ship less JavaScript: render content on the server and keep only genuinely interactive components on the client.
• Load third-party scripts — analytics, chat widgets, tag managers — after the page is interactive, and remove the ones you don't need.
• Break up long tasks by yielding to the main thread between chunks of work.
• Import only what you use from large libraries such as icon packs, date libraries and charting libraries.

Step 4: Eliminate Cumulative Layout Shift (CLS)

• Give every image, video and iframe explicit width and height attributes (or a CSS aspect-ratio) so the browser reserves space.
• Reserve space for ads, embeds, cookie banners and anything else injected after load.
• Use size-adjusted fallback fonts so swapping in the web font doesn't move text.
• Animate with transform and opacity instead of properties that change layout, such as top or height.

Step 5: Measure, change one thing, and measure again

Lab scores vary by a few points between runs because of network and CPU noise, so test several times and compare the median. Change one thing at a time so you know what helped.

SpeedAudit AI audits every page in your sitemap on mobile and desktop, tracks scores over time, and uses Gemini AI to turn each failing audit into a prioritised fix — so you can see which change moved the score and catch regressions early.