A developer writing on DEV Community describes the order in which they fix Core Web Vitals problems on client marketing sites. Core Web Vitals are three measures of page experience: Largest Contentful Paint (LCP), which is how fast the main content appears; Interaction to Next Paint (INP), which is how quickly the page reacts to a tap or click; and Cumulative Layout Shift (CLS), which is how much the layout jumps while loading.
The first step is to measure. The author runs PageSpeed Insights on key pages such as the homepage, a service page and an article. They look at field data first, because it reflects real visitors, and treat lab data as a second opinion from one simulated visit. They record all three metrics so they have a baseline.
The second step is the largest image, which is often the main cause of poor LCP. The author checks whether the file is much bigger than the space it fills, whether it uses a modern format such as WebP or AVIF, whether it has width and height attributes, and whether it is wrongly lazy-loaded. Hero images should not be lazy-loaded and should be preloaded or given high fetch priority.
The third step is removing what blocks the first paint, such as large CSS files, web fonts and scripts in the head. The fourth is reserving space for images, embeds and banners so the layout stops shifting. The last is INP, where the advice is to remove, defer, then simplify heavy JavaScript. The author also says to avoid chasing a perfect score, to compare several test runs, and to be wary of adding a plugin to fix a plugin problem. If fixes are not enough, the cause may be hosting, caching, the database or the theme.
Why it matters
Slow sites are hard to fix when nobody knows which part is slow. Splitting the problem into three metrics, each with its own usual cause, turns a vague complaint into a short checklist. The suggested order also puts the cheapest, most effective work first, which suits small teams with limited time.
The advice on restraint is useful too. Scores vary between runs, and a perfect lab result is not the same as a better experience for visitors. Looking at field data and repeating tests helps you avoid spending effort on changes that make no real difference. Note that the post also promotes the author's own services, so treat that part as marketing.
Source: DEV Community. This brief summarises their reporting; read the original for the full detail.