High TTFB
The browser waits too long before receiving the first byte.
Web performance
Is the website slow?
Before optimizing, you need to know where time is being lost. A server-side problem is not fixed by compressing images, and a busy main thread is not fixed by adding more cache.
Where slowness can come from
Poor performance does not always mean the same thing. These signals help separate server, network and browser work.
Slow LCP
The main content appears late because of images, fonts, CSS or the initial response.
Delayed interaction
JavaScript and long tasks keep the main thread busy.
Heavy resources
Images, video, fonts or bundles transfer more data than necessary.
Weak caching
Assets or application responses are regenerated and downloaded too often.
Third-party scripts
Tags, widgets and external services add work and dependencies.
How I find the bottleneck
I measure before changing the site and compare every intervention with a baseline.
Baseline
Collect TTFB, waterfall, Core Web Vitals and transfer size under repeatable conditions.
Separate causes
Distinguish backend, network, frontend and third parties to avoid working on the wrong layer.
Prioritize
Start from the changes that materially affect loading and interaction.
Verify
Compare results after each change with the initial data.
Monitor over time
When useful, combine lab tests with real-user data.
What I measure
Different stacks need different tools. The diagnosis is built around the actual site.
TTFB and waterfall
LCP, INP and CLS
JavaScript and long tasks
Blocking CSS
Images and fonts
Caching and CDN
Backend and database
Third-party scripts
Goal
Reduce unnecessary work, not chase a score
The useful outcome is a site that responds better under real conditions. Synthetic scores help with diagnosis, but they are not the end product.
- Measured server response
- Optimized key resources
- Reduced JavaScript where needed
- Consistent caching
- Monitoring of useful metrics
Technical references
Want to know what is actually slowing the site down?
We can start from a technical baseline and identify where it makes sense to work first.