Website Animation: Types, Tools and Performance Without Compromise
Website animation done right: micro-interactions, scroll animation, Lottie, CSS, GSAP and WebGL, plus how to protect speed, Core Web Vitals and accessibility.
Read
Core Web Vitals are three Google metrics for user experience: LCP (how fast the main content loads – good is 2.5 s or less), INP (how quickly the page responds to input – 200 ms or less) and CLS (visual stability – 0.1 or less). They are assessed at the 75th percentile of real visits, separately for mobile and desktop. To speed up your site, start with field data in Search Console, find the metric that is failing – usually LCP on mobile – and fix it at template level rather than page by page.
Core Web Vitals are the subset of Web Vitals that Google considers essential for every page. Each covers a different part of the experience: loading, interactivity and visual stability. Results are assessed at the 75th percentile – a page passes a metric if at least three out of four visits fall within the threshold, separately on mobile and desktop (web.dev).
| Metric | What it measures | Good | Needs improvement | Poor |
|---|---|---|---|---|
| LCP (Largest Contentful Paint) | Time until the largest image or text block in the viewport is rendered | ≤ 2.5 s | 2.5–4 s | > 4 s |
| INP (Interaction to Next Paint) | Time from a click, tap or key press to the next frame; the longest interaction counts, ignoring outliers | ≤ 200 ms | 200–500 ms | > 500 ms |
| CLS (Cumulative Layout Shift) | Total of unexpected layout shifts in the worst “session window” (max 5 s) | ≤ 0.1 | 0.1–0.25 | > 0.25 |
INP replaced the older FID metric on 12 March 2024. It is a big shift: FID measured only the delay of the first interaction, whereas INP observes every click, tap and key press throughout the visit. Scrolling and hovering are not counted.
Google does not disclose how much weight they carry. Its Search Central documentation says good Core Web Vitals, along with other page experience aspects, align with what its core ranking systems seek to reward, and it explicitly recommends that site owners achieve them.
In practice, treat CWV as a hygiene factor rather than a lever that will push weak content to the top. A fast page that misses the searcher’s intent will not beat a slower but clearly more helpful one. Speed also matters beyond rankings: a slow page loses visitors whose clicks you may already have paid for in ads.
First, distinguish between two kinds of data. Field data comes from real Chrome users and is what the assessment is based on. Lab data is a simulated single load on a defined device – great for diagnosis, but it is not the result Google sees.
| Tool | Data type | What it is for |
|---|---|---|
| Core Web Vitals report in Search Console | Field (CrUX) | Groups of similar URLs with problems; validating fixes |
| PageSpeed Insights | Field and lab | Quick overview of a single URL and the whole origin |
| Chrome DevTools (Performance panel) | Lab, compared against field data | Diagnosis: what delays LCP, which interactions are slow, what shifts the layout |
| Lighthouse | Lab | Technical audit; cannot measure INP, so it shows TBT instead |
| CrUX (Chrome UX Report) | Field | Historical results and origin comparisons |
The Search Console report groups URLs with a similar user experience, and a group’s status is set by its slowest metric: poor CLS with LCP that needs improvement means the group is “poor”. If a group lacks enough data, its URLs may be left out of the report. On low-traffic sites you are then left with lab data and your own real-user monitoring (RUM).
LCP is the most common problem. According to the Web Almanac 2025 (July 2025 data), 62% of sites have good LCP on phones, compared with 77% for INP and 81% for CLS. On 73% of mobile pages the LCP element is an image, as web.dev’s round-up reports, so start there.
| LCP subpart | What it covers | Recommended share |
|---|---|---|
| TTFB | Time to the first byte of HTML from the server | ~40% |
| Resource load delay | Time from TTFB until the LCP image starts downloading | under 10% |
| Resource load duration | Downloading the image itself | ~40% |
| Element render delay | Time from download to the element appearing on screen | under 10% |
These proportions come from web.dev’s guide to optimising LCP and are guidelines, not hard rules. If resource load delay is eating a whole second, the problem is not the image’s file size – it is that the browser finds out about the image too late.
<img> with src or srcset. According to web.dev, 35% of LCP images had URLs that were not discoverable in the initial HTML, and 7% of pages hide them in a data-src attribute – typical of lazy-loading plugins.fetchpriority="high" to the LCP image. Only 15% of eligible pages do this. Do not give high priority to more than one or two images, or the effect gets diluted.loading="lazy" from the hero image.<head> and unnecessary CSS. Render content on the server rather than waiting for JavaScript.font-display value other than auto or block, so text does not wait for the font file.Be wary of hero carousels. An image that only appears once the slider script runs is, by definition, discovered late.
Every interaction has three phases (web.dev): input delay – the time before event handlers start running; processing duration – running the code; and presentation delay – the time until the next frame is painted. The usual culprit is a busy main thread.
scheduler.yield() lets you hand control back to the browser mid-task – it works in Chrome and Edge 129+ and Firefox 142+, but not in Safari, so you need a fallback such as setTimeout.content-visibility to off-screen sections.Lighthouse cannot measure INP because it does not perform real interactions – it reports TBT as a proxy. To diagnose slow interactions, use the DevTools Performance panel while clicking through the page as a user would, or rely on field data.
According to web.dev, the typical causes are images and videos without dimensions, web fonts that render larger or smaller than their fallbacks, ads and widgets that resize after loading, and content injected above existing content. Shifts within 500 ms of a click or key press are excluded, because the user expects them.
width and height attributes or aspect-ratio for images and video. 66% of pages have at least one unsized image.min-height) for ads, embeds and anything loaded asynchronously.transform and opacity, not top, left or margin. Pages that animate layout-affecting properties are 15% less likely to have good CLS.Cache-Control: no-store header and unload listeners disable the instant back/forward restore that eliminates shifts.Core Web Vitals are only one part of technical SEO. For the full checklist, see our step-by-step SEO audit checklist.
WordPress sites and page builders tend to repeat the same patterns:
It is easier to design a new site for speed from day one than to rescue it after launch. That is how we approach website design and development and e-commerce builds, where one slow product template is repeated across thousands of URLs. If you are planning rich animation or 3D graphics, read how to balance them with performance in our article on 3D websites built with WebGL.
Is your site failing Core Web Vitals? We will find what is slowing it down and fix the causes, not the symptoms, as part of our SEO and GEO service. Get in touch and send us your URL.
Good means LCP of 2.5 s or less, INP of 200 ms or less and CLS of 0.1 or less, measured at the 75th percentile of visits, separately for mobile and desktop. Poor starts above 4 s for LCP, 500 ms for INP and 0.25 for CLS. Anything in between needs improvement.
FID measured only the delay of the first interaction on a page. INP observes every click, tap and key press during a visit and covers the whole time until the next frame is painted, not just the initial delay. INP replaced FID as a Core Web Vital on 12 March 2024.
The 0–100 score in PageSpeed Insights is a lab test of one simulated page load. Search Console shows only field data from real users, aggregated for groups of similar URLs. The Core Web Vitals assessment is based on field data, so that is the data that matters more.
Field data comes from real visits, so changes do not appear immediately. Once you start validation, Search Console runs a 28-day monitoring session. Meanwhile, you can track progress in PageSpeed Insights or with your own real-user monitoring.
Google recommends good Core Web Vitals and says they align with what its core ranking systems seek to reward, but it does not disclose how much weight they carry. Speed will not replace relevant, helpful content. Treat CWV as a baseline for a good site, not a shortcut to the top spot.
Lighthouse loads the page without real user interactions, so it cannot measure INP. It reports Total Blocking Time (TBT) as a proxy instead. You can diagnose slow interactions in the Chrome DevTools Performance panel or from field data.
Insights

Website animation done right: micro-interactions, scroll animation, Lottie, CSS, GSAP and WebGL, plus how to protect speed, Core Web Vitals and accessibility.
Read
Elementor vs a custom WordPress theme: speed, SEO, security, cost and editing. When a page builder is enough, and what leaving Elementor did for our own site.
Read
How to write a website brief: a template to copy covering goals, audience, competitors, content, features, SEO, integrations, budget and timing, plus mistakes.
ReadWe design and code company websites from scratch — no off-the-shelf templates, built for conversion, speed and visibility in Google and AI search.
We use essential cookies and, with your consent, analytics (Google Analytics, Hotjar) and marketing cookies to understand how you use the site and improve it. Details in our privacy policy.