See which page element costs your users the most time — the server, scripts, images or fonts — and get a fix order you can compare after shipping (Lighthouse, CrUX, waterfall).
Interactive demo
A guided walkthrough of the module view: score, tabs, report chapters and Core Web Vitals from the sample scenario — no sign-in.
Hover to choose how to start
No sign-in — take the tour or click through the demo modules on your own.
Want the full Insight demo? Take the full app tour — dashboard, audit and report.
That red line on the waterfall is your LCP. Instead of a bare score you get the concrete element, the request and the fix order (Lighthouse, CrUX field data, waterfall, asset analysis).
Then it groups issues, estimates impact and orders the fixes — instead of leaving you with a score alone.
Lab, CrUX or no measurement — never faked.
The metric, asset or request that actually costs time.
A fix priority and a retest after shipping.
Four steps from test to decision — without guessing where to start.
LCP, INP, CLS — lab data and field data when available.
Heavy scripts, long tasks and interaction impact.
Images, fonts, CSS and third-party network requests.
Measure key metrics, server and assets for a specific URL.
Connect a weak score to an element, request or code fragment.
Start with changes that have the highest impact and a sensible cost.
Compare before and after under the same measurement conditions.
Metrics and context
The goal is improving a specific user metric without making other areas worse. The module separates field data (CrUX), lab measurement (Lighthouse / Playwright) and Insight’s synthetic score — and when data is missing, it shows why instead of a zero.
Code and assets
JavaScript, CSS and media are the most common sources of delay. The module shows which scripts block the main thread, which styles delay rendering and which files can be deferred or optimized.
Largest payload · srcset / WebP
Load process
The waterfall shows DNS, connection/TLS, TTFB, download, main-thread blocking and render-blocking resources. The filmstrip shows what visitors actually see — not just test numbers.
Mobile and history
You compare profiles within the same run or historical measurements of the same URL and page version. A weaker CPU and slower connection often reveal issues invisible on desktop.
cometweb.io/Our own site, measured before and after the fix on 28–29 September 2026.
Read the full case studyPageSpeed Insights, mobile · TBT from 2.7–5.4 s to 0.10–0.26 s
No. Lighthouse runs a lab test under controlled conditions. Core Web Vitals may also come from real-user data (CrUX) when available. The CometWeb module score is Insight’s own scale — we do not copy Lighthouse colour bands 1:1.
A quick test is a public single-URL check without an account. A Performance audit in Insight is the full module in a project context: history, subpages, export and a retest after shipping.
A lab load test has no real user interaction — so we use TBT instead of INP. In field data, INP appears only with enough CrUX samples; missing data is N/A with a reason, not zero.
No. Insight analyzes the page, surfaces issues and helps you build a work plan. Implementation still requires changes in code, CMS, server config or third-party services.
Not for basic analysis of a public URL. Additional setup may be needed for selected integrations, monitoring or non-public pages.
Analyze a specific URL, find what slows loading and start with the highest-impact fix.