A classifier that trusts the provider's own bot field first and falls back to user-agent patterns for prefetchers the provider has not classified. Flagged events are stored with is_machine set rather than discarded.
Also called: Apple Mail Privacy · Gmail image proxy · fake open rates · did they really read it
- 1An open event arrives with client info.
- 2The provider's bot field is checked first and is authoritative.
- 3If empty, user agent and client name are pattern-matched for known proxies and scanners.
- 4The event row records is_machine so reporting can separate human from machine opens.
The route explains the inflation and the specific wrong fix that preceded it: 'Apple Mail Privacy Protection fetches that image on Apple's servers for every message its users receive, read or not... a number that quietly overstates itself is worse than no number.' And on the user-agent trap: 'Gmail's image proxy identifies itself as ... an ordinary desktop browser string, with "bot": "gmail" alongside it. Matching on the string logged every Gmail recipient as a human open one second after send, which is the exact inflation this whole module exists to avoid.'
- Privacy proxies inflated open rates to the point of meaninglessness.
- User-agent sniffing alone misclassified an entire mail provider's recipients as human.
See it on your own jobs
Twenty minutes, your numbers, no slide deck. We’ll build one of your real buildings in front of you and send you the estimate link at the end — yours to keep either way.
or keep browsing features →