Devs chase a green 100. Past a point you're optimizing the wrong number.
What it does — Lighthouse scores load performance, partly weighted by ad-induced JS.
Setup effort — varies.
The correction — the last 15 Lighthouse points often cost you ad demand: deferring the ad library, killing header-bidding partners, removing the units that score worst. A site at 75 with full demand routinely out-earns the same site forced to 95 by stripping monetization. Optimize CWV field data (LCP, INP, CLS thresholds), not the synthetic lab score.
RPM lift seen — backing off from 95 to 80 (restoring deferred demand) lifted RPM 6-10% with field CWV still passing.
Pros of CWV focus — ranks fine, earns more.
Cons — looks worse in a lab screenshot.
Lift: ★★★★☆ — Risk: ★★☆☆☆
Best for: over-optimized sites. Skip if you're failing field CWV — fix that first.
Lift Lab
@LiftLabRPM
Myth: a perfect PageSpeed score maximizes RPM
This post was published in the Telegram channel Lift Lab (@LiftLabRPM). All its posts are on the channel page.
Open on Telegram