When interpreting PSI (PageSpeed Insights), first distinguish whether content appears too slowly, interactions respond too slowly, or the layout is unstable. Then prioritize improvements using the relevant metrics rather than focusing solely on the overall score. Drawing on articles from 2022 to 2024, this guide summarizes how to interpret the results and approach optimization. References to Core Web Vitals versions and recommended values retain their historical context and are not treated as current platform rules.
Check each device type, then review user experience and performance diagnostics
PSI is a free web page testing tool provided by Google. According to Welly’s 2024 article, reports are divided into mobile and desktop, each offering sections to understand real-user experience and diagnose performance issues. Both device types are worth checking: good desktop results do not necessarily mean browsing on a phone will be equally smooth.
The article identifies LCP, INP, and CLS as the Core Web Vitals for 2024, covering the loading of the largest content element, interaction responsiveness, and visual stability, respectively. Earlier articles still list FID as a core metric. When reading older guides, note that INP replaced FID in that update; lists from different periods should not be used interchangeably.
LCP and FCP: content appearing does not mean the main content is ready
FCP measures the time until the first text or image appears; LCP measures the time until the largest content element is rendered. For example, if text appears while a large hero image is still loading, the two metrics describe different stages. Da Vinci’s 2022 article recommends an FCP below 1.8 seconds and an LCP below 2.5 seconds. These are reference values given in that article.
- The largest content element loads slowly: Check large image file sizes, compress images, and streamline CSS, JavaScript, and page structure.
- The first content takes too long to appear: Check for render-blocking resources, server response delays, and multiple redirects, and consider caching static resources.
- Adjust the loading order: Preload critical resources and lazy-load secondary content, rather than lazy-loading every image. While custom fonts are still loading, display a fallback font to keep text visible.
INP: check more than the first click
FID focuses on the first input, whereas INP covers interaction responsiveness throughout a visit. When checking buttons and keyboard interactions, therefore, do not assess only whether the first interaction is smooth. Da Vinci’s 2022 article recommends an INP below 200 milliseconds and suggests the following improvements:
- Remove unused CSS, JavaScript, and third-party code.
- Break up code execution and defer loading less important JavaScript.
- Simplify the DOM—the page’s document structure—by reducing unnecessary nodes.
CLS: do buttons suddenly move as content loads?
CLS measures layout stability. For example, if you are about to click a button when a loading image pushes it away, that is a situation worth investigating. Da Vinci’s 2022 article recommends a CLS below 0.1. Its suggested improvements include setting image and video dimensions in advance or reserving space with containers, avoiding automatically appearing content that pushes the layout around, and reducing animations that cause layout shifts. These issues call for layout adjustments, not just file compression.
Interpreting and improving PSI metrics: perspectives from across the web
What can you improve when TTFB is high?
TTFB is the time it takes for a browser to receive the first byte from the server. If the wait is long, check hosting performance, bandwidth, and application code, and consider using a CDN or reducing multiple redirects according to the site’s needs. Da Vinci’s 2022 article recommends keeping TTFB below 800 milliseconds. This is a reference value from that article and should not be treated as a current platform rule.
What additional information does Speed Index (SI) provide?
SI focuses on how quickly content becomes visually apparent on screen, helping assess what users experience while waiting for a page to appear. If the results are poor, Da Vinci’s 2022 article recommends reducing main-thread and JavaScript work and ensuring that text remains visible while fonts load. The article uses a value below 3.4 seconds as a reference for fast performance.
Does a higher PSI score prove that a website is better overall?
No. A PSI score alone cannot tell you how good a website is overall. Da Vinci’s article considers a high score an indication that a site meets Google’s criteria for a good website. Welly’s article, however, notes that some large websites may still offer a good user experience despite lower scores and advises against focusing too heavily on achieving a perfect score. A more reliable approach is to review individual metrics alongside diagnostic recommendations, identify specific problems, and address them.
References
- PageSpeed Insights (PSI) 是什麼? 對SEO有影響?www.da-vinci.com.tw · 2022-09-30T12:47:58+08:00 — Discusses: What can you improve when TTFB is high?; What additional information does Speed Index (SI) provide?
- PageSpeed是什麼?會影響SEO嗎?網站測速工具這裡看!welly.tw · 2024-10-07T10:41:12.274Z — Discusses: What can you improve when TTFB is high?; Does a higher PSI score prove that a website is better overall?




