The Challenge: Suboptimal LCP on Key Pages
Monrif is an Italian media company operating major publications like Il Resto del Carlino, Il Giorno, and La Nazione. With millions of users and thousands of articles, navigation speed is critical for reader retention.
Before optimization, LCP values on article pages ranged from 2,333ms to 2,419ms - limiting perceived speed when users navigated between articles.
Goal: Improve user retention by making navigation feel instant.
The Solution: Speculation Rules API
The team implemented the Speculation Rules API to prerender article pages before users click on them. This modern API replaces older prefetch methods with more precise control.
Setting Moderate Eagerness
The prerendering trigger was set to "moderate" level:
``
html
``
Technical Exclusions
The script was configured to ignore:
This gave precise control over what should be preloaded.
Results Achieved (Desktop)
| Publication | LCP Improvement | Engagement Increase |
|-------------|-----------------|---------------------|
| Il Resto del Carlino | -17.9% | +6.9% |
| Il Giorno | -17.1% | +6.7% |
| La Nazione | -15.0% | +5.3% |
Peak engagement increase: 8.9% during periods of highest prerender activity (June 2023).
Correlation Between Performance and Engagement
The data showed clear correlation:
Additional Benefits: Advertising
Instant Back Navigation with bfcache
The team removed "unload" event blockers and implemented Maverick's Permissions-Policy to enable back/forward cache (bfcache). This allowed instant restoration of pages from browser memory.
Higher Ad Viewability
Average ad viewability on opened pages improved to 1.03%, increasing advertiser confidence and quality of ad inventory.
Faster Creative Loading
Ad creatives loaded within 500ms increased by 10.4% on prerendered desktop pages.
Key Takeaways
---
> Source Attribution: This content is adapted from the case study "How Monrif improved engagement by 8.9% and reduced LCP by 17.9% with Speculation Rules prerender and back/forward cache" published on web.dev under the Creative Commons Attribution 4.0 License.