Performance website checks
Website performance checks examine resource delivery and the browser work needed to make a page useful. Large images, unnecessary scripts, blocking requests and ineffective caching can delay a visit. Laboratory observations describe specific conditions and cannot replace field data. A fast homepage does not establish that a product page or checkout performs equally well.
Start with the resource or element behind the delay. Resize images to their displayed role, reserve space for media and defer work the first view does not need. Cache versioned assets without caching personalised HTML. Repeat measurements under comparable conditions and review slower devices before declaring a site-wide improvement. Separate loading, stability and interaction metrics.
The check pages below describe the exact observable condition, why it matters, practical fixes and the limits of the detector. Open the affected URL and inspect its evidence before making a change. After correction, repeat the scan and verify the result in a browser. The related guide connects these individual checks into a review workflow. Keep deliberate exceptions documented so that a passing automated result is never mistaken for a complete manual assessment.
Checks and fixes
Cache-Control Header Missing: Declare the Right Policy
Fix a missing Cache-Control header. Choose policies for HTML, versioned assets and private responses, verify revalidation, and avoid unsafe shared caching.
Cumulative Layout Shift Too High: Find and Stabilise Movement
Fix high Cumulative Layout Shift. Locate moving elements, reserve image and embed space, tune fonts, and compare the scan with standard CLS and field data.
First Contentful Paint Too Slow: Reduce the Blank Screen
Improve slow First Contentful Paint. Diagnose server delay, blocking CSS, scripts and fonts, test mobile rendering, and compare lab results with field data.
Gzip compression not enabled on text responses
Fix missing text compression. Enable negotiated gzip or Brotli, inspect Content-Encoding, test HTML and assets, and avoid streaming or double-encoding failures.
HTML Document Size Too Large: Reduce the Initial Payload
Reduce oversized HTML without hiding useful content. Find inline data and repeated markup, distinguish transfer from decoded size, and test content discovery.
Images Not Served as WebP or AVIF: Verify Actual Formats
Review images not served as WebP or AVIF. Check actual response types, convert suitable assets, use responsive sources, and recognise URL-based scan limits.
Largest Contentful Paint Too Slow: Diagnose the Critical Path
Improve slow Largest Contentful Paint. Identify the LCP element, separate server and image delays, prioritise resources, and compare lab with field data.
Slow Time to First Byte: Separate Network and Server Delay
Improve slow time to first byte. Separate network from server work, profile expensive routes, cache safely, and compare scanner and browser timing.
Too Many HTTP Requests: Audit the Work Behind the Count
Reduce unnecessary HTTP requests. Audit third-party chains, defer noncritical media, preserve consent and core features, and verify bytes and main-thread work.
Too Many Inline Scripts: Separate Code from Data
Review too many inline scripts. Separate executable code from JSON-LD and state, reduce repeated work, preserve load order, and verify actual blocking time.
Too Many Render-Blocking Stylesheets: Reduce Critical CSS Work
Audit too many render-blocking stylesheets. Remove unused CSS, shorten discovery chains, preserve the cascade, and verify real blocking rather than link count.
Total Blocking Time Too High: Reduce Long Main-Thread Tasks
Reduce high Total Blocking Time. Profile long tasks, cut unused JavaScript, split costly work, and distinguish the loading metric from real interaction latency.