Website & speed audit: what to check, what each score means, and what to fix first
Only 55.9% of the world's websites pass all Core Web Vitals, and the LCP threshold tightened from 2.5s to 2.0s in 2026. The three metrics, their pitfalls, the field + lab methodology, and the highest-impact fixes — prioritized by impact × effort, not a raw list of errors.
A website & speed audit is the technical analysis of a site across three axes — performance (Core Web Vitals), technical health (crawl, indexing, security) and experience (mobile, accessibility) — that ends with a list of fixes prioritized by impact × effort, not a raw list of errors.
The goal isn't the "Lighthouse score," but the speed perceived by the real user and the revenue you lose when the page is slow. This study shows what to check, what each metric means, and what to fix first.
- Only 55.9% of sites pass all Core Web Vitals (CrUX, May 2026). The rest lose ranking and conversion.
- 2026 thresholds:LCP < 2.0s (tightened from 2.5s), INP < 200ms, CLS < 0.1 — judged on real users, the 75th percentile.
- INP is the hardest to fix (~43% fail): it demands less JavaScript on the main thread, not just an optimized image.
- Field + lab, then impact × effort. You don't fix everything — you fix first what moves the needle.
- The deliverable is a decision, not a PDF. A score per dimension + the top 10 prioritized fixes + an estimate of the gain.
Why it matters in 2026
Speed is no longer just a comfort: it's a ranking and conversion factor. Per CrUX (May 2026), only 55.9% of the world's websites pass all three Core Web Vitals — the rest lose out. And the threshold has become stricter: in the March 2026 Google update, LCP tightened from 2.5s to 2.0s. Sites that pass every metric have, on average, a ~24% lower bounce rate and a measurable ranking advantage.
Key point to understand: Google evaluates at the 75th percentile of real users (field data, not lab data). In other words, at least 75% of visitors must get a "good" score on each metric. A green Lighthouse on your laptop guarantees nothing if your real users are on mobile over 4G.
The three Core Web Vitals
| Metric | What it measures | “Good” | The pitfall |
|---|---|---|---|
| LCP — Largest Contentful Paint | load speed of the main element | < 2.0s | heavy hero images, slow server, no preload |
| INP — Interaction to Next Paint | responsiveness to interaction | < 200 ms | heavy JavaScript on the main thread |
| CLS — Cumulative Layout Shift | visual stability | < 0.1 | images/ads without width/height |
INP is the metric most sites fail — ~43% miss the 200ms threshold in 2026. It replaced FID (First Input Delay) in 2024 and is the one that demands the deepest technical changes (cutting and splitting JavaScript).
What a full audit checks
- 01
Performance (Core Web Vitals + beyond)
LCP, INP, CLS on field data (CrUX) and lab data (Lighthouse); TTFB (server response), page weight, number of requests, render-blocking resources, image strategy (WebP/AVIF, lazy-load, explicit dimensions), fonts (preload + font-display: swap).
- 02
Technical health / technical SEO
Crawlability (robots.txt, sitemap.xml), indexing (indexed vs. orphan pages, canonicals, duplicates), internal link structure, redirects (chains, 404s), JS rendering (critical content visible without JS — essential for AI crawlers too), structured data.
- 03
Security & infrastructure
HTTPS and certificate hygiene, security headers (CSP, HSTS), outdated versions/dependencies, exposed surfaces, known vulnerabilities.
- 04
Mobile & accessibility
Mobile-first (the majority of traffic), tap targets, viewport, contrast, accessibility (WCAG) — which helps both SEO and conversion.
How it's done (methodology)
A proper audit combines field data (CrUX / Search Console — what real users experience) with lab data (Lighthouse / WebPageTest — to isolate the cause), then prioritizes by impact × effort. You don't fix everything — you fix first what moves the needle.
The four highest-impact fixes for LCP: preload the hero image, inline critical CSS, preload fonts with display: swap, and server-side rendering / cache. For CLS: explicit width/height on every image, video, iframe and ad slot. For INP: less JavaScript on the main thread, splitting long tasks, selective hydration.
Monitoring tip: set alerts at 80% of the thresholds (INP > 160ms, LCP > 2.0s, CLS > 0.08), so you catch degradation before you fail the official threshold.
Common findings (what we find almost every time)
Hero images weighing several MB with no preload (bad LCP); JavaScript from tag managers and third parties that blocks interaction (bad INP); cookie banners and ads with no reserved dimensions (bad CLS); old WordPress installs with dozens of plugins that pile on requests; critical content rendered client-side only (partly invisible to crawlers).
What you get from an audit (the deliverable)
Not a list of errors, but: a score per dimension (Overall / Performance / Technical / Security), the top 10 fixes prioritized by impact × effort, and an estimate of the gain (speed, ranking, conversion). In other words, a plan for months 1–3, not an 80-page PDF.
Întrebări frecvente
Does a Lighthouse score of 100 mean the site is fast?
Not necessarily. Lighthouse is a lab test, run on a simulated configuration. Google judges on real field data (CrUX, 75th percentile). A site can score 95 in Lighthouse and still fail Core Web Vitals for real users on mobile.
Which metric is the hardest to fix?
INP. It requires cutting JavaScript and restructuring how the page responds to interaction — deeper changes than optimizing a single image for LCP.
Does speed really affect ranking?
Yes, but indirectly and alongside content quality. Core Web Vitals are a ranking signal, and good speed lowers bounce rate (~24%) and lifts conversion — the business effect is often larger than the pure ranking effect.
How long does an audit take?
A quick diagnostic in a few days; a full audit with a prioritized plan in 1–2 weeks, depending on the size of the site.
Surse folosite în studiu
The adoption and threshold data come from Google (CrUX, core update); the rest from 2026 industry guides, flagged as such.
- 01Google · CrUXMay 2026
Chrome User Experience Report
Only 55.9% of sites pass all three Core Web Vitals. Evaluated at the 75th percentile of real users (field data).
- 02Google SearchMar. 2026
Core update — LCP threshold
The “good” LCP threshold tightened from 2.5s to 2.0s.
- 03linkDigital Applied2026
Core Web Vitals 2026 — INP, LCP, CLS optimization guide
2026 thresholds, INP as the metric most sites fail (~43%), and the highest-impact fixes.
- 04linkcorewebvitals.io2026
Core Web Vitals — metrics reference
Definitions and thresholds for LCP, INP and CLS.
Want the audit done by us, with a prioritized plan?
A diagnostic that shows you where you're losing speed, ranking and conversion — plus the top 10 fixes by impact × effort. No strings attached.