Page Speed & Core Web Vitals

How fast a page loads and how smoothly it responds are part of how search engines judge page experience — and independently, they decide whether people stay long enough to read anything. Google publishes a specific set of metrics for this, the Core Web Vitals.

These thresholds are published by Google, which is why they can be quoted precisely — unlike most ranking specifics. INP replaced an earlier metric, First Input Delay, in 2024; older material still referring to FID is out of date.

Poor LCP is nearly always a large image or a slow server response. The single most common cause is an uncompressed hero image at the top of the page.

Poor INP is JavaScript. When the main thread is busy executing scripts, it cannot respond to a tap — the page looks loaded but does not react. Large frameworks, heavy third-party scripts, and analytics or chat widgets are common contributors.

Poor CLS is content appearing and pushing other content down: images without width and height, ads or embeds inserted after load, and web fonts swapping in at a different size than the fallback.

1. Compress and resize images. Almost always the largest single win, and the easiest.
2. Set width and height on every image. Fixes CLS directly, costs nothing.
3. Remove third-party scripts you do not need. Each one is someone else's code executing on your page, on their schedule. Audit what is actually there — sites accumulate these.
4. Lazy-load below-the-fold images, but never the top one.
5. Serve modern image formats — WebP or AVIF over JPEG and PNG.
6. Enable caching and compression on the server. Usually a configuration change rather than a code change.

Two kinds of data, and the difference matters:

• Lab data — a single test run in controlled conditions. Lighthouse in browser developer tools and PageSpeed Insights both provide this. Useful for diagnosis, because it tells you exactly what is slow.
• Field data — measurements from real visitors on real devices and connections. Search Console's Core Web Vitals report shows this. It is what actually reflects your visitors' experience.

A page can score well in a lab test and poorly in the field, usually because real users are on slower devices and networks than the machine running the test.