[CASE STUDY] 100/97 Lighthouse on High‑Traffic News Portal WITHOUT CDN — Pure LSCache + LSWS Optimization

[WebPrestige]

Hello everyone,
I would like to share some performance results from a WordPress news site (protinews.com.gr) running on LiteSpeed Web Server with LSCache. The site uses the Newspaper theme and receives heavy daily traffic with constant updates. No CDN is used at the moment.
I am posting this mainly to discuss the LSCache configuration and server tuning that helped us reach very good Lighthouse scores on both desktop and mobile.
Desktop results (March 2026): LCP 1.9s, INP 16ms, TBT 20ms, CLS 0.05, Speed Index 0.7s, Performance score 100.
Mobile results: LCP 2.4s, INP 87ms, TBT 0ms, CLS 0.00, Speed Index 1.8s, Performance score 97.
These numbers were achieved without Cloudflare or any external CDN. Only LSCache and server‑level optimization were used.
Some of the techniques applied:

• public cache for all visitors
• private cache for logged‑in users
• ESI for dynamic blocks
• guest mode enabled
• selective JS defer
• HTML/CSS/JS optimization
• Brotli compression
• HTTP/3 enabled
• PHP LSAPI tuning
• reduced DOM size
• removed heavy theme modules

The goal was to see how far LSCache can go on its own before adding QUIC.cloud. . The next step will be enabling QUIC.cloud CDN to improve TTFB and global delivery.
I am sharing this to exchange ideas with the community. If anyone has suggestions for further improvements or wants more details about the configuration, I would be happy to discuss.


Happy to answer questions or share more configuration details if needed.

 

[serpent_driver]

A PageSpeed score has nothing to with "Speed of Page" or performance or good or bad the LSWS configuration is. PageSpeed measures how fast the content is displayed AFTER the loading assets are displayed in browser window or in other words, PageSpeed measures how fast a page "feels". PageSpeed (and page optimization) is just browser cosmetic. Learn more about this narrative: https://www.cachecrawler.com/Blog/W...nce-Doesnt-Start-With-Optimization::6601.html

Or more uncomfortable truths about PageSpeed and optimization. https://www.cachecrawler.com/Blog:::42.html

 

[WebPrestige]

Thanks for your reply. I understand your point about PageSpeed. In my case the site is on shared hosting, so I can’t adjust any server settings. The only area I can work on is the frontend, and that’s why I’m checking LCP, CLS and INP. These affect the user experience directly. If you have any tips for improving this under shared hosting limitations, I’d appreciate it.

 

[serpent_driver]

Just invest some minutes and read the linked pitch carefully. At the end LiteCache Rush will solve all performance issues!

https://www.cachecrawler.com/Rush:::45.html

 

[hellen-charless]

Great results, thanks for sharing this with the community.
Those Lighthouse scores are impressive, especially considering the site is running without a CDN and still achieving such strong mobile and desktop performance. It really shows how effective server-level tuning can be when properly configured on LiteSpeed Web Server with LiteSpeed Cache.
The combination you mentioned—public cache, private cache for logged-in users, and ESI for dynamic elements—is often overlooked but can make a huge difference on high-traffic news sites. Enabling Guest Mode and selective JS defer is also a smart approach for improving Core Web Vitals without breaking theme functionality.
Running a site on WordPress with the Newspaper Theme can sometimes lead to large DOM sizes and heavy scripts, so removing unused modules was likely a key factor in achieving those numbers.
When you move to the next step with QUIC.cloud, you’ll probably see improvements in TTFB and global latency, especially for visitors outside your primary server region. At that stage, it may also be worth testing:

• Critical CSS generation through QUIC.cloud
• Image optimization and WebP/AVIF delivery
• Edge caching for anonymous users
• Object caching (Redis or Memcached) if not already enabled

Overall, it’s great to see a real-world example showing how far LSCache can go with proper tuning. It would be interesting to know the server specs (CPU, RAM, PHP version) and whether you’re using any object caching layer, as that could help others replicate similar results.
Thanks again for sharing the setup and performance data!

 

[serpent_driver]

• Critical CSS generation through QUIC.cloud
• Image optimization and WebP/AVIF delivery
• Edge caching for anonymous users
• Object caching (Redis or Memcached) if not already enabled

You're following a typical and widespread narrative. PageSpeed and Lighthouse cannot measure actual loading time and performance. Therefore, server performance measures play no role in PageSpeed and Lighthouse and do not change the score.

 

[WebPrestige]

Thanks a lot for the detailed and thoughtful feedback, really appreciate it.
You’re absolutely right about the impact of removing unused Newspaper modules — the theme can easily generate a very large DOM and several heavy scripts if everything is enabled. Cleaning that up, combined with ESI for dynamic blocks and Guest Mode, made a noticeable difference on a high‑traffic news site where content updates constantly.
Regarding QUIC.cloud, that’s exactly the next step. Once DNS access is finalized, the plan is to enable the CDN so we can improve TTFB and global delivery, especially for visitors outside Greece. At that point I’ll also test Critical CSS generation and the image optimization pipeline (WebP/AVIF), since both should help stabilize LCP even further during peak traffic hours.
For context, the site is currently on standard shared hosting (PHP 8.2 LSAPI, no Redis/Memcached available), so LSCache is doing most of the heavy lifting. That’s why I wanted to share the results — to show what can be achieved even without root access or a CDN, just by tuning LSCache properly and optimizing the frontend.
Happy to share more configuration details or specific LSCache rules if it’s useful for others in similar setups.