Core Web Vitals Aren't a Developer Problem — They're a Design Problem

"We'll optimize performance after launch" is one of the most expensive sentences in web design, because by launch, most of what determines performance has already been decided — in the design file, not the codebase. Google's Core Web Vitals — the metrics it uses to grade real-world page experience — reward or punish decisions made long before a single line of code gets written.

Largest Contentful Paint (LCP) measures how long the biggest visible element takes to render, and needs to land under 2.5 seconds to be considered good. Interaction to Next Paint (INP) measures how responsive the page feels when someone actually clicks or taps something, with a good threshold under 200 milliseconds. Cumulative Layout Shift (CLS) measures how much content unexpectedly jumps around as the page loads, and should stay under 0.1. All three are measured at the 75th percentile of real visitors, not a lab test on a fast laptop — meaning the experience for someone on a mid-range phone on patchy mobile data is exactly the experience being graded.

A hero section with a full-bleed, uncompressed image as the largest visible element is a design decision that directly sets the LCP number before any developer touches it. A custom webfont loaded without a fallback strategy causes visible text reflow the instant it swaps in — that's CLS, decided by a typography choice. A page that auto-injects an ad slot, a cookie banner, or a promotional bar above existing content without reserving space for it is the single most common cause of layout shift on the web, and it's a layout decision, not a technical one. Engineers can mitigate a bad design decision here and there; they can't fully undo one baked into the visual hierarchy of the page.

Core Web Vitals matter beyond a PageSpeed score for two concrete reasons: Google has used page experience as a ranking input since 2021, and slow, jumpy pages measurably cost conversions — visitors abandon a page the longer it takes to become usable, which is exactly what LCP and INP are trying to quantify. Treating performance as a design constraint from the first mock-up, not a cleanup task before launch, is the difference between hitting these thresholds and chasing them for months afterward.