Core Web Vitals are the specific performance metrics Google factors into ranking, and they're worth understanding individually rather than just chasing a single Lighthouse score — because each one has a different, specific cause.
Largest Contentful Paint (LCP) — how fast the main content appears
Measures when the largest visible element (usually a hero image or heading) finishes rendering. On a typical Framer site, LCP is commonly held back by: an unoptimized hero image (wrong format, oversized for its display dimensions), font loading from an external CDN adding a connection round-trip before text can render in the correct typeface, or render-blocking JavaScript delaying paint. Good target: under 2.5 seconds.
Cumulative Layout Shift (CLS) — how much the page jumps around while loading
Measures unexpected layout movement as a page loads. Framer sites are usually reasonably disciplined here since the editor generally reserves space for elements — but a common culprit when CLS is bad is images or embeds without explicit dimensions, causing surrounding content to jump once the resource finishes loading and its actual size is known. Good target: under 0.1.
Interaction to Next Paint (INP) — how responsive the page feels to actual clicks/taps
Measures the delay between a user interaction and the page visibly responding. This is where Framer's runtime weight shows up most directly — a heavier JavaScript bundle means more main-thread work competing with your click/tap handler, which can measurably delay the page's response even on a device that's not obviously struggling. Good target: under 200ms.
What a runtime-stripped export typically improves
Removing Framer's JS runtime directly helps INP (less competing main-thread work) and often LCP (less render-blocking script), though FNJ's exports don't do that: all three keep Framer's runtime, so most of the original JavaScript still loads. Self-hosting images and fonts helps LCP specifically; FNJ's exports self-host images as WebP, while fonts still load from Framer's CDN. CLS is largely unaffected by export itself, since it's mostly a content-authoring concern rather than a platform overhead one — if your original Framer site had CLS issues, they typically carry through unless explicitly addressed.
How to actually check yours
Don't rely on lab data alone (Lighthouse) — check PageSpeed Insights for real-world "field data" from actual visitors (the CrUX report), where available for your domain. Lab and field data can diverge, and field data is what actually factors into ranking. If your site doesn't have enough traffic for field data yet, lab data from Lighthouse is the next best signal, run on your actual live URL rather than a local preview.
Related conversion check
If your goal is lighter pages, compare the original with FNJ's Framer to HTML output on the same deployed host and device class. Keep the result that improves your real page without losing critical interactions.
Starting a new site instead? FNJ's AI website builder designs one from a description and a site you like, then gives you the code to host anywhere.
