It is more secure (though not end-to-end secure, of course) from the PoV of most threat models because the monitoring most are most concerned about is at their side of the connection (“local infra (often wireless) -> through ISP -> …” or “local -> through ISP -> through state/org level monitoring -> …”).

Tracking you by monitoring the network traffic between CF (or other connection aggregator) and the origins is possible, but usually significantly more effort. Those with reason to be concerned about true end-to-end encryption will know that seeing any CF artefacts means they need to investigate to know if they are getting e2ee or just end-to-proxy.